Download The Role of Subjectivity in Intelligent Systems Communication and

Document related concepts

Philosophy of artificial intelligence wikipedia , lookup

AI winter wikipedia , lookup

Agent-based model wikipedia , lookup

Personal knowledge base wikipedia , lookup

Incomplete Nature wikipedia , lookup

Upper ontology wikipedia , lookup

Agent-based model in biology wikipedia , lookup

Ethics of artificial intelligence wikipedia , lookup

Agent (The Matrix) wikipedia , lookup

Knowledge representation and reasoning wikipedia , lookup

History of artificial intelligence wikipedia , lookup

Cognitive model wikipedia , lookup

Embodied cognitive science wikipedia , lookup

Renzo Gobbin
A Doctoral Thesis submitted in fulfilment of the
requirements for the degree of
Doctor of Information Sciences by Research
University of Canberra
School of Information Sciences and Engineering
September 2011
© Renzo Gobbin - All Rights Reserved
The role of subjectivity in intelligent systems communication and
by Renzo Gobbin
This thesis examines the role of subjectivity in intelligent systems communication interfaces.
It explores new models of software agent architectures able to provide agents with
subjectivity and egocentric communication patterns.
A qualitative approach has been utilised using disciplines such as Intelligent Systems,
Philosophy, and Cognitive Sciences.
The research results aim to increase knowledge of the subjective intelligent systems
communication internal processes modelling, so improving the design of human-intelligent
agent interface tools.
This research may be applied to intelligent agent and knowledge systems adopting subjective
and egocentric communication in areas such as analysis of security intelligence, legal
argument analysis and intelligent web search engines.
Table of Contents
Abstract ______________________________________________________ ii
Form B ______________________________________________________ iii
Table of Contents______________________________________________ iv
List of figures _________________________________________________ vi
Acknowledgments ____________________________________________ viii
Chapter 1: Introduction__________________________________________ 1
1.1 Introduction & Motivation_______________________________________ 1
1.2 Research Problem ______________________________________________ 5
1.3 Background ___________________________________________________ 6
1.4 Research Questions ____________________________________________ 11
1.5 Research Objectives ___________________________________________ 12
1.6 Research Relevance ___________________________________________ 13
1.7 Theoretical Stance_____________________________________________ 15
1.8 Research Methodology _________________________________________ 17
1.9 Thesis organisation and layout __________________________________ 19
Chapter 2: Literature Review ____________________________________ 21
2.1 Literature review introduction __________________________________ 21
2.2 Why subjective communication is an important research topic________ 23
2.3 Software Agents communication overview. What is known? __________ 25
2.4 Mediated communication activity ________________________________ 31
2.5 Current issues in software agent modelling and architectures _________ 34
2.6 Agents learning, cognition and consciousness ______________________ 36
2.7 Subjective communication, what is unknown on this research area ____ 36
2.8 The knowledge gap we are trying to fill ___________________________ 38
2.9 Chapter Summary ____________________________________________ 40
Chapter 3: ActivityTheory and Intelligent Systems Communication _____ 41
3.1 Intelligence and Tool Mediated Activity. __________________________ 41
3.2 The development of concepts in artificial intelligent systems __________ 43
3.3 Intelligent tool mediated communication activities __________________ 45
3.4 Development of Artificial Intelligent Systems Knowledge ____________ 49
3.5 Chapter Summary ____________________________________________ 50
Chapter 4: Subjective Communication Issues _______________________ 51
4.1 Subjective Intelligent Communication ____________________________ 51
4.2 Intelligent Agents Objective and Subjective properties integration ____ 57
4.3 The role of identity in intelligent exchanges ________________________ 58
4.4 Chapter Summary ____________________________________________ 60
Chapter 5: Subjective Intelligent Systems Ontologies Applications ______ 62
5.1 Formal and applied ontology ____________________________________ 62
5.2 ALMA Ontology Matching Techniques ___________________________ 67
5.3 Legal Intelligent Subjective Systems Application Ontology ___________ 69
5.4 LKIF Legal Ontology framework ________________________________ 71
5.5 Chapter Summary ____________________________________________ 72
Chapter 6: ALMA Subjective Intelligent System Test Model ___________ 73
6.1 ALMA multi-agent architecture environment description and set-up __ 73
6.2 ALMA Model Artefact Architecture______________________________ 78
6.3 Legal ontology conversion to Jade Java beans ontology ______________ 82
6.4 Chapter Summary ____________________________________________ 89
Chapter 7 ALMA Model Subjective Communication Analysis __________ 91
7.1 ALMA Single Subjective Multi-agent Cluster Model ________________ 91
7.2 ALMA Single Subjective Cluster A1 Data Analysis _________________ 93
7.3 ALMA Single Subjective Cluster A2 Data Analysis _________________ 98
7.4 ALMA Double Subjective Multi-agent Cluster Model ______________ 101
7.5 ALMA Double Subjective Cluster B1 Data Analysis________________ 103
7.6 ALMA Double Subjective Cluster B2 Data Analysis________________ 112
7.7 Externalisation of subjective properties analysis ___________________ 121
7.8 Chapter Summary ___________________________________________ 123
Chapter 8: Conclusions ________________________________________ 126
8.1 Research Background_________________________________________ 126
8.2 Research Conclusions _________________________________________ 129
8.3 Research Contribution ________________________________________ 131
8.4 Future Research _____________________________________________ 133
Bibliography_________________________________________________ 135
Appendix A: List of Publications ________________________________ 147
Appendix B: Protégé Ontology Editor ____________________________ 148
Appendix C: JADE Agent Development Environment _______________ 151
Appendix D: LKIF JADE ALMA Classes _________________________ 158
Appendix E: Eclipse Java IDE __________________________________ 183
Appendix F: Agent ALMA and Cogito Ontology Code Java Code ______ 209
Appendix G: PiersonvPost Legal Case ____________________________ 214
Glossary ____________________________________________________ 223
List of figures
FIG. 1 MULTIDISCIPLINARY RESEARCH LAYOUT ..........................................................................5
FIG. 2 THESIS STRUCTURE..................................................................................................................20
FIG. 3 SUBJECTIVE MULTI-AGENTS TAXONOMY .........................................................................27
FIG. 4 ACTIVITY THEORY MODEL.....................................................................................................46
FIG. 5 MULTIPLE AGENTS TOOL MEDIATED ACTIVITY MODEL. ..............................................47
FIG. 6 ONTOLOGY TAXONOMY (JAQUETTE 2002) .........................................................................63
FIG. 7 AGENT ONTOLOGY COMMUNICATION ...............................................................................68
FIG. 8 LEGAL ONTOLOGY RESEARCH STRUCTURE......................................................................79
FIG. 9 ALMA EXPERIMENT STRUCTURE .........................................................................................80
FIG. 10 ALMA AGENT CLUSTER MODEL..........................................................................................81
FIG. 11 PROTÉGÉ ONTOLOGY EDITOR .............................................................................................83
FIG. 12 PROTÉGÉ ONTOLOGY NETBEANS CONVERTER ..............................................................84
FIG. 13 COGITO ONTOLOGY IN PROTÉGÉ FORMAT......................................................................85
FIG. 14 OBJECTIVE AGENT PROCESS................................................................................................87
FIG. 15 ALMA AGENT PROCESS .........................................................................................................88
FIG. 16 SUBJECTIVE AGENT MODEL.................................................................................................89
FIG. 17 ALMA SUBJECTIVE MULTI AGENT MODEL ......................................................................92
FIG. 18 JADE SNIFFER MESSAGES STRUCTURE EXPERIMENTS A1-A2.....................................93
FIG. 19 DA0-OBJECTIVE- ALMA AGENT EXCHANGES..................................................................94
FIG. 20 ALMA AGENT FIRST EGOCENTRIC COMMUNICATION..................................................95
FIG. 21 ALMA-SELF-SUBJECTIVE AGENTS EXCHANGE ...............................................................96
FIG. 22 SUBJECTIVE- DA0 AGENT EXCHANGE...............................................................................97
FIG. 23 DA0-OBJECTIVE MESSAGE EXCHANGE .............................................................................98
FIG. 24 EGOCENTRIC COMMUNICATION PROCESS.......................................................................99
FIG. 25 EXTERNALISATION COMMUNICATION PROCESS.........................................................100
FIG. 26 EXPERIMENT B META-AGENTS ARCHITECTURE ..........................................................102
FIG. 27 EXPERIMENT B1 DIAGRAM.................................................................................................104
FIG. 28 EXPERIMENT B1 COMMUNICATION MODEL HISTORY................................................104
FIG. 29 EXPERIMENT B1 FIRST INTERNALISATION PROCESS ..................................................105
FIG. 30 FIRST B1 EGOCENTRIC MESSAGE EXCHANGE...............................................................106
FIG. 31 TEST B1 EGOCENTRIC MESSAGE.......................................................................................107
FIG. 32 ALMA2 INTERNALISATION PROCESS...............................................................................108
FIG. 33 ALMA2 ONTOLOGICAL SHIFT PROCESS ..........................................................................109
FIG. 34 ALMA2 EXTERNALISATION PROCESS..............................................................................110
FIG. 35 AGENTS SUB2 EXTERNALISE TO OBJ AND DA1.............................................................111
FIG. 36 EXPERIMENT B2 DIAGRAM.................................................................................................112
FIG. 37 MODEL B2 COMMUNICATION FLOW................................................................................113
FIG. 38 MODEL B2 STARTING QUERY ............................................................................................114
FIG. 39 MODEL B2 ALMA CLUSTER INTERNALISATION PROCESS..........................................115
FIG. 40 MODEL B2 ALMA CLUSTER EGOCENTRIC COMMUNICATION...................................116
FIG. 41 MODEL B2 EXTERNALISATION PROCESS........................................................................117
FIG. 42 AGENT OBJ2 INTERNALISATION PROCESS .....................................................................118
FIG. 43 AGENT ALMA2 EGOCENTRIC VALIDATION PROCESS .................................................119
FIG. 44 MODEL B2 AGENT SUBJ2 EXTERNALISATION PROCESS .............................................120
FIG. 45 SUBJECTIVITY MATRIX .......................................................................................................122
FIG. 46 PROTÉGÉ ONTOLOGY EDITOR ...........................................................................................149
FIG. 47 ONTOLOGY BEANS GENERATOR PLUG-IN......................................................................150
FIG. 48 JADE AGENT DEVELOPMENT ENVIRONMENT ...............................................................153
FIG. 49 JADE DF REGISTER WINDOW .............................................................................................154
FIG. 50 JADE DUMMY AGENT MESSAGE FORM...........................................................................155
FIG. 51 JADE SNIFFER AGENT WINDOW ........................................................................................156
FIG. 52 JADE INTROSPECTOR AGENT WINDOW ..........................................................................156
FIG. 53 JADE PLATFORM....................................................................................................................157
FIG. 54 JADE ALMA LEGAL ONTOLOGY ........................................................................................158
I wish to thank the University of Canberra School of Information Sciences & Engineering
and the Dean of School Professors Dharmendra Sharma, for the assistance given during the
research period 2000-2007. I wish also to sincerely thank Professor Craig McDonald for his
appreciated scientific stimulation and valuable academic suggestions on critical times.
Particular thanks to my supervisor Associate Professor Masoud Mohammadian for the
academic and personal support given during the research period. Finally I wish to thank my
wife Maria for her infinite patience and support during the whole research period.
Chapter 1: Introduction
1.1 Introduction & Motivation
The Masters thesis dissertation “ The role of cultural fitness in user resistance to
information technology tools” (Gobbin 1998a), chapter two, related to philosophical
grounding of IT tools and Human Computer Interaction, emphasised the importance of
subject-object relationships and the use of Vygotsky’s cognitive theories as vital for IT
tool mediated activity analysis. During masters research time, Bonnie Nardi’s seminal
book “Context and Consciousness” (Nardi, 1996) aroused my interest and subsequently I
discussed the subject with HCI mediated activity scholars such as Bonnie Nardi (US),
Kary Kuutti (Finland) and Victor Kaptelinin (Sweden). These exchanges and
developments in Intelligent Systems Human Computer Interaction field greatly increased
my interest in this field complemented by work experience with Xerox Corporation
Artificial Intelligence from 1982 to 1995 provided me with an insight that Activity Theory
could address Intelligent Systems communication issues involving subjectivity. In my
Masters Thesis conclusion I enclosed a note stressing the importance of further research
in the area of Vygotsky Activity Theory in particular for applications in intelligent system
communication and learning.
Disciplines in which I had majored at the Australian National University such as
philosophy of mind, anthropology, cognitive science and activity theory practised while
working in artificial intelligence systems provided further motivation in pursuing a
multidisciplinary research investigation. Subsequently, advances in Intelligent Systems
software engineering technology, Intelligent Software Agents and in communication and
coordination in the Internet environment in the area of business intelligence and AI and
Law applications, provided me with the motivation to start a PhD research project on
Intelligent Software Agent models and architectures able to incorporate subjectiveobjective mediated communication activity theories relevant to Human-software agents
interactions applying Vygotsky’s tool mediated activity theories. The concept of agency
can be construed as the product of high-order cognitive processes or in its usual
philosophical usage as first-order phenomenal experience (Gallagher 2007). In this thesis
we are using an activity theory perspective where the concept of agency is constantly
constrained by social grouping, material and symbolic resources as well as cognitive
In activity theory perspective agency is a negotiated relationship between the subject and
the tools used by a subject to perform activities in a determinate context (Bernat 2011).
Seen from the activity theory perspective Agency is in fact liable to change in response to
new contextual developments, feature that is important in software agent communication
activities. In this context, the common use of software agents “agency properties” is in
terms of an intelligent tool able to achieve operational goals. Artificial intelligence
techniques employed to create intelligent software tools performing autonomous tasks
have been currently developed in ways that may attempt to display some of the features
related to intelligent entities in the animal world. If we examine current intelligent
software agent applications we see that they are used in a metaphoric “intelligent tool”
way aiding human users to perform high level intellectual activities. A software agent
essential function is to act as effective bridge between a person's goals and expectations
and the computer's capabilities. The agent-tool metaphor is used to make the interface
more intuitive and to encourage types of interactions that might be difficult to evoke with
a standard interface. Agents of this sort do not need to be explicitly anthropomorphic
although this is the arena in which the expressive qualities of 'agents characters' such as
belief, desire and intentions (BDI) are being explored and start to be developed.
The recent advances in researching intelligent multiple agent cooperation and
communication require a deeper understanding of the internalisation - externalisation
dynamics between thought and speech on the human user side while it is interfacing with
inferential knowledge and agent communication language used in multiple agents system
environment coordination. Goal driven cooperative activities mediated by the use of
tools are deeply ingrained in human behaviour and have existed since the emergence of
hunting and gathering societies. The complexity of modern global enterprise working
environments requires communicative interaction with intelligent systems able to interact
with humans in a meaningful way (Lock & Peters 1996).
Externalisation of internal conscious knowledge is an important expressive quality that is
embedded in human cooperation, communication, learning and knowledge management
activities. Researching the subjective externalisation and the internalisation of objective
information united with objects identification in a space/time domain ontology is an
important step forward for an adequate analysis and design of intelligent systems
architectures enabling intelligent user interfaces able to be adaptive, and easily
incorporated in organisational working tools and cultures.(Gobbin 1998b, 1998c)
Modern e-commerce and enterprise management activities are heavily based on human
cooperative activities mediated by complex and powerful ICT communication tools
requiring an adequate integration of both the engineering and the user communication
models. The complexity of modern ICT communication and business environments has
generated an enormous amount of business activity over the Internet between public and
corporate entities and at the same time companies and corporations are increasing their
use of Information agents and Knowledge Management systems for rapid decision
making and market policy implementation (Klush 2000).
The use of classic Human-Computer Interaction (HCI) theories such as Vygotsky’s
Activity Theory (Kaptelinin 1996; Kuutti 1996) could be very important for the analysis
of subjective communication exchanges between humans and software agents in view of
establishing a new human-intelligent software agent interaction model where human
users could communicate and exchange knowledge and conceptual data with artificial
intelligent systems. Subjective intelligence implies the externalisation of a degree of
reasoning and learned behaviour where a software agent has the ability to subjectively
accept/refuse a user’s statement of goals and tasks carrying out the delegated tasks by
using a series of subjective communication exchanges. Knowledge is what we gain when
we perceive things and we often use our intellect to come to conclusions about these
perceptions. Intellect is our ability to take this knowledge and come to conclusions with it
through reason. Essentially our intellect is how good we solve problems and is objective.
The only problem is that intellect is dependent on our knowledge, what we perceive and
conclude, for us to present it in some fashion. Our knowledge is definitely subjective so
our representation of intellect is also doomed to be subjective in some manner; at least at
this point in time.
To address the issue of subjective communication exchanges between human users and
intelligent software agents, we require the analysis and design of new Intelligent Software
Agents (ISA) architectures models. These models should be able to provide objective
internalisation with learning and retention of conceptual knowledge data and subjective
externalisation of internal knowledge the agent posses by subjective communication
exchanges. Multiple agents need to use a range of communicative tools in order to
transfer messages in a context of intelligent agent cooperative activities. Cooperation is
hereby defined as the activity of multiple agents working together requiring agents’
communicative capacity in order to exchange information about their common goals,
their own identity, and their current status (Franklin & Gaesser 1996; Odell 2000; Cohen
& Levesque 1995).
Software agent communication technology has recently been utilised to form clusters of
cooperative agency activities enabling such agents to produce and receive ACL SpeechAct messages. Although deprived of the richness of human verbal communication, an
embryonic form of speech-act communication is achieved. The same internalisation and
externalisation processes of human speech and language could then be modelled in
communicative agents’ internal processes that could also be externalized in sets of
communicative behaviour using specific agent languages as a mediating tool (Finin, et al.
1994; FIPA 1997, 2002).
This PhD research thesis will attempt to model cooperative multiple software agents with
the following characteristics:
1. Subjective and Objective qualities that are required by intelligent agents to perform
bi-directional multiple communication activities.
2. The ability to internalise representations of perceived communicative patterns.
3. The ability to externalise internally stored representations of communication
patterns to other agents.
In Fig. 1 a layout showing the multidisciplinary aspects of the research model will provide
an introduction to the topics that we will examine later on. An intelligent software agent
model with the characteristics described above could provide a novel paradigm for
multiple agent interactive cooperation activities adopting a multidisciplinary approach.
The use of human language tools mediated activity theories proposed by Vygotsky for
the explanation of cognitive development will provide also a strong theoretical
framework for a thoroughly investigation of multiple cooperative intelligent agent
communication activities (Vygotsky 1986, 1978).
1.2 Research Problem
An analysis of IT Human-Computer Interaction design practices shows a continuing
evolution (Grudin 1990) from hardware to software and from software to a higher level
of cognitive processes that is now extended also into complex organisational processes
with related business intelligence interfaces. Computer application tools are in fact
providing an extension to, and sometimes incorporating pre-computer human activities
involving the use of tools (Grudin & Gentner 1996; Kaptelinin 1996). The design and use
of IT tools is currently following cultural patterns similar to a variety of other tools used
by humans and the design of software agents or softbots tends to follow traditional tool
design practices. However, when we are designing intelligent systems artefacts able to use
semantic tools themselves in communication exchanges with humans and other software
agents, the communication process that is taking place involves the internalisation and
externalisation of semantic concepts (Gobbin 1998b).
Current system analysis and design methodologies are following standardised methods
for the interpretation and description of productive sequential human activities. These
methodologies could be insufficient to map the rich semantic cooperative activities
among intelligent systems exchanging conceptual knowledge as they are not taking into
account subjectivity. Intelligent software agent communicative acts require adequate
models, architectures and design methodologies able to capture the subjectivity of
internalized knowledge communication patterns. In addition to the analysis of objective
intelligent agents’ external communicative activities, we should attempt to model the
“internal areas” of an agent communicative activities with the aim of determining the role
that subjectivity can have in the areas of intelligent agent communication and learning.
This includes the necessity of creating an internal domain ontology that can assist in the
capture, internalisation, storage and externalization of subjective conceptual knowledge.
This research thesis will attempt to address the above problems by recognizing the
symmetric pattern of communication exchanges between intelligent systems and try to
define a new experimental intelligent system model addressing these problems. This
research can contribute to improving the design and modelling process of intelligent
agent systems able to internalise and subjectively externalise domain knowledge
improving human-intelligent systems interaction and communication.
1.3 Background
The questions posed in researching artificial intelligent systems communication are
similar to those arising when researching human communication. We should note here
that intelligent communication activities are not mere translation where, from a given
input, an output is required. Quite often software agent communicative activities are
misunderstood as encoding/decoding processes related to the interpretation of complex
linguistic codes. The work of an encoding/decoding device is not certainly inferential or
creative (Sperber 1994). It is not inferential because the symmetric relation between a
message and a signal is different from the asymmetric relation of premises to conclusion
i.e. the meaning of a sentence does not logically follow from its encoded physical
expression of a sound. Also an encoding/decoding device is rarely creative. It would be a
quite challenging device if it was changing form during communication because a creative
encoding/decoding device that changes the process symmetry quite often will generate
communication errors at every symmetry change (ibid. 1994). Encoding/decoding can
simply be an ancillary communication process or an interfacing tool while intelligent
systems cooperative communication activities are more complex than encoding/decoding
as software agents produce and receive Agent Communication Language (ACL) speechact semantic messages (FIPA 1997, 2002). Although ACL is deprived of the richness of
human verbal communication an embryonic form of semantic communication is still
achieved. The same internalisation and externalisation processes of speech and language
can be modelled in communicative agents’ internal processes that are also externalized
and internalized in sets of communicative behaviour using standard agent languages as
mediating tools.
In order to model intelligent agent subjective properties, intelligent software agent
architectures must include:
1. Language mediated communicative activities;
2. Subjective and Objective properties required by intelligent agents to perform
subjective communication activities;
3. The ability to internalise representations of communicative knowledge patterns and
subsequently to externalise internally stored knowledge representations to other
agents or humans.
The subjective software agent characteristics described above can form a platform for
modelling intelligent agent interactive cooperation activities by applying tools mediated
Activity Theory proposed by Vygotsky for the explanation of cognitive development and
learning thus providing an appropriate theoretical framework for investigating intelligent
systems communication activities models and for the design of novel intelligent
architectures integrating Activity Theory (Vygotsky 1978, 1986).
Activity Theory places subjective and objective dialectic as being central to the process of
learning and cognition by using language as a mediating tool in the cognitive
internalisation process. Subjectivity has been one of the more important philosophical
investigations since Aristotle and it is still actively researched by modern philosophers
(Chalmers 2000).
When discussing Subjectivity and Objectivity in the following notes we will be using
philosophical concepts applicable to artificial to intelligent agents:
1. Objectivity refers to the view that the truth of a thing is independent from the
observing Subject (in our case an intelligent agent or a person). This notion entails
things that exist independently from an agent or are external to the agent itself.
Objective truths are independent from an Agent’s internal beliefs and intentions.
In our investigation the concept of Objectivity is important for communicative
internalization activities and the necessity for intelligent agent internal models;
2. Subjectivity on the other end denotes that the truth of some class of externalised
statements (Verbal or not) depends on the internal states or reactions of an
intelligent entity (agent) making the statement. The notion of Subjectivity is that
knowledge is restricted to an agent’s own perceptions and the object qualities
experienced by the agent are subject to its interpretations (Strawson 1959).
In contemporary philosophy of science and epistemology, scholars employ the notion of
Objective/Subjective distinction (Strawson 1956; Evans 1982; Strawson 2008) while in
philosophy of mind it is a matter of controversy whether the notion of subjectivity is
epistemic, metaphysical or both (Nagel 1974). In my investigation of intelligent agent
subjectivity models I left out the metaphysical question where something is objective in
the case where it exists independently from the agent and subjective otherwise. I have
found it more appropriate to apply Strawson and Evans’ theories for intelligent agent
technology where internal states’ independence is at play in distinctions between the
Objective and Subjective status of an agent. Strawson’s theories give a description of various
concepts that form an interconnected web, representing (part of) our common, shared,
human conceptual scheme. In particular, he examines the conceptions of basic
particulars, and how they are brought under the area of general spatio-temporal concepts.
What makes this a metaphysical project is that it shows, in fine detail, the structural
features of our thought about the world, and thinking about reality. Evans’ theories also
involved conceptual representation descriptive reference arguing that causal antecedents
of the information involved in a mental state are claimed to be sufficient to determine
which object the state concerns (Strawson 2008; Evans 1982; Odemberg 2009). Cognitive
science, philosophy and in particular phenomenology have tried to explain the higher
metaphysical properties of intelligence based on subjectivity and consciousness. Recent
advances in neuroscience research are trying to solve the problem of consciousness that,
while one of the main targets of philosophical enquiry, recently has been accepted as
scientific research on the quest to find how matter can produce phenomenological
conscious states and self representation (Edelman 2000; Damasio 2000; Revonsuo 2006).
Communicative intelligent agents should perform both Subjective and Objective roles in
order to perform communicative activities. In this context an intelligent software agent
must be able to communicate and, for this reason, represent itself as a Subject in an
Objective world whose existence is independent of the agent itself. An agent’s ability to
represent itself as an Object among other agents in an Objective spatial environment will be
used in investigating models for an intelligent agent architecture that will include the
concept of an agent-self. Peter Strawson’s theories on descriptive metaphysics (Strawson
1966) and spatial representation theories applied in recent research on self representation
(Ricoeur 1992) together with cognitive robotic self representation will be applied also in
our intelligent agent self representation research. In the following paragraphs I provide
introductory notes on Intelligent Systems to act as a path leading to the research thesis
questions. As Intelligent Systems are involved and the research thesis will focus on
Intelligent Software Agents, we will start now to briefly describe Intelligence and then we
will discuss how Intelligent Systems are able to use properties related to intelligence for a
number of Artificial Intelligence applications.
Since ancient times, intelligence was considered as the ability to read internalised concepts
and to rationally link and select what was considered true knowledge and essential
meanings of conceptual reality. By considering the etymology of the term intelligence we
can also see that the term intelligence is derived from the Latin composite “Inter-Lego”
which morphed into “Inteligo” that literally means “to read inside” and also “to select
from many things inside those that are true”. (Vico 1988) However, even today the
actual issue of categorisation of intelligence remains quite open as the tasks involved
include environmental, cultural and social variables. Also models of learning,
internalisation and externalisation of intelligence are continuously under investigation.
While Piaget has offered an explanation of children’s intellectual development as an
evolving movement from sensory motor activity to higher thinking levels, (Piaget 1954)
Vygotsky proposed a phenomenological concept based on culture, learning and
socialisation. Higher thinking functions according to Vygotsky are accelerated by the
active “mediating” use of physical or mental tools such as language, culture and nonverbal social communication (Vygotsky 1978). Capability and use of linguistic tools are
very important for the development of higher mental states, egocentric thought and
consciousness. The highest degrees of intelligent behaviour in the animal world involve
the use of some kind of tool (Parker 1993).
Current intelligence and intellectual development research is basically oriented towards
humans and high level animal species such as primates. However, advances in areas such
as Artificial Intelligence and Intelligent Systems together with recent developments in
World Wide Web communication technology and Natural Language interfaces have
increased the number of intelligent systems applications where some aspects of
intelligence and intelligent behaviour are required.
From the point of view of Artificial Intelligence and Intelligent Systems intelligent
properties are used in the following activities:
Thinking as humans (Turing Test approach);
Acting and communicating as humans (Simulation);
Rational analysis (Logic and rule based thinking);
Rational action (Rational software agents).
Artificial intelligence mainly includes theories and techniques for the development of
algorithms and software providing computing machinery with the ability of simulating
specific intelligent human activities in a specified environmental domain (Luger &
Stubblefield 1998). Recent developments in intelligent software agent communication
involve the ability to communicate with other agents by using mediated tools such as
Agent Communication Language (ACL) and semantic tools such as natural language. The
communicative properties of intelligent software agents and the active use of mediated
linguistic tools will be important for scientific research in intelligent behaviour simulation
and the possible use of agents for intelligent systems and humans interfaces. The ability
of an intelligent agent to simulate, express or externalise humanlike intelligence and
phenomenal conscious traits will require a more comprehensive architectural model than
the current autonomous software agent model (Gobbin 2006).
A truly communicative intelligent software agent should in fact be able to develop and
learn by internalising data from the world and consciously building internal phenomenal
conceptual knowledge that can be externalised in the communication process with
humans and other agents. These processes require both a) an adequate architectural
model making tool mediated communication activities possible, and b) a theoretical
framework that will produce self constructed conceptual entities and internal knowledge
The research background described above constitutes a preamble for the introduction of
questions about how Intelligent Systems such as intelligent software agent architectures
could be designed in such a way as to be able to internalize and externalize knowledge
and represent themselves as subjects in an objective world. The achievement of such an
intelligent system model will enable the analysis of artificial intelligent artefacts
internalised knowledge and, by paraphrasing Thomas Nagel’s paper “What is it like to be
a Bat?”, (Nagel 1974) perhaps to be able to explain “What it is like to be an intelligent
software agent?” by exploring an intelligent system architecture’s internal communication
1.4 Research Questions
There should be a degree of inter-relationship between an intelligent entity self experience
awareness and a subjective phenomenal dimension of the environment where it operates.
Also the notion of the self implies an identity theory able to identify the self in time and
space as different from other entities (Dennet 1991; Wegner 2002; Metzinger 2003).
Philosophical phenomenology theories can offer much in the area of artificial intelligent
system consciousness and also in the area of self-identity. It certainly addresses issues
crucial to an understanding of consciousness and also offers a conceptual framework to
understanding subjectivity (Zahavi 2005).The principal reason behind Vygotsky’s research
on subjectivity and developmental learning was that subjectivity is a central concern for
phenomenological disciplines including the self, consciousness and intentionality.
Vygotsky’s scholarly studies prior to his developmental psychology breakthroughs were
based on Hegel and Husserl’s phenomenology. Vygotsky’s Activity Theory proposes that
communication, learning and the egocentric use of language as mediating tools are
important to develop and build internal higher mental states with phenomenological
content (Vygotsky 1978).
The use of philosophy and phenomenology in the design of intelligent systems
architectures simulating aspects of human intelligence is also validated by recent
neuroscience research addressing organisational models of the human mind. A complete
model of the human mind will be incomplete unless the phenomenal mechanism of
consciousness, an important part of internal intelligent behaviour, is included in the
model. A theoretical model of the mind will require access to empirical data for all mind
levels (including phenomenological ones) and neuroscience is not yet able to provide a
methodology necessary to build from neuronal physical data a complete model that
includes consciousness (Revonsuo 2006). The reason for this problem is the actual
impossibility of reading empirical data from the physical brain and at the same time
model a conceptual mapping referring to the analysed empirical data. External attempts
to create an internal model will always be in the third person while an internal first person
modelling methodology is necessary.
As Intelligent Systems are designed and built to attempt simulations of human
intelligence models, we can see that incomplete human mind models may present
problems in modelling Intelligent Systems simulations of internal states. These problems
may be addressed by the integration of phenomenology, consciousness and theories of
the self in the analysis of Intelligent System architectures. The possibility of researching
and experimenting with intelligent systems simulating human intelligent behaviour and
the building of models including consciousness and self identity could provide intelligent
system architectures that integrate phenomenological aspects of intelligence. By using a
meta-structured intelligent agent network we will be able to capture internal empirical
data between multiple areas of the intelligent model under research so that
phenomenological internal aspects of the intelligent system conceptual knowledge can be
properly analysed. Such an architectural model could then provide answers to the
following research questions:
1. What is the role that subjectivity could play in artificial intelligent systems
communication and learning?
2. Can the intelligent system model internal environment architecture achieve
subjective internalisation and egocentrism?
I should clarify here that any unintelligible correlation between phenomenal objective
data and the artificial intelligent system internal architecture could not count as the
explanation of consciousness because the aim of this research is to find out how internal
phenomena may work and how they could be modelled by monitoring the intelligent
meta-structure internal communication patterns between internal structure modules. A
model of artificial intelligent system phenomenal conscious properties is related to how
the intelligent system under research will model the internalisation of objective reality.
The questions above present research objectives examined in the next section and are
related to the investigation and construction of an intelligent software agent metastructure model capable to provide the necessary functions and methodology.
1.5 Research Objectives
The objectives of the research will need to answer the questions above i.e. the role that
subjective externalisation has in intelligent software agent communication. The research
will involve a number of complementary disciplines although the main objectives can be
narrowed to:
1. Create a Subjective Intelligent Systems Model: - Abstract description of intelligent
software agent communication behavioural activity with subjective intentional
2. Design Subjective Agents Architectures: - design and implement a software architecture
where software agents with intelligent subjective properties could operate;
3. Verify Subjective Intelligent Systems Theories: - Theoretical verification of subjective
theories such as AI, Intelligent systems, Cognitive Sciences, Learning theories,
philosophy of mind, philosophy of language, applied ontology, Neuroscience;
4. Application of Subjective Intelligent Software Agents in mission critical working environments:
Business Intelligence, e-Commerce, e-Business, e-Health, On-Line Disputes
Resolutions, Human-Computer Communication, Knowledge Management, IT
Service Management. (Configuration Management).
1.6 Research Relevance
The current research will address the problem of robustly modelling cooperative software
agent communicative acts using Activity Theory (AT) when performing business
intelligence and knowledge mining cooperation’s in enterprise environments. Rather than
concentrating on objective agents external communicative activities, I am investigating a
model able to show internal communicative activities with the aim of determining the
role that subjectivity has in the areas of cooperation, internal agent architecture, agent
internal knowledge and self identification in the environment. This includes inter alia the
modelling of communicative context ontologies, internal knowledge storage and
reorganization of knowledge in real time or in multiplex mode. While system engineering
research is active in physical network agent cooperation and communication in the
Internet environment, a multidisciplinary study of agent internalization of knowledge
using subjective models has marginally been attempted. I believe that the research will
enhance domain knowledge in the area of agent identification and re-identification of
objects external to the agent self and therefore improve the introspection process and
agent concept of the self. The research will also be extremely important in the area of
Internet knowledge mining and eCommerce where many agents can communicate and
collaborate for the achievement of common goals. Being able to identify private goals
against other agents behaviour will then facilitate and expedite knowledge mining and
processing. European and US research on Intelligent Agents are mainly involved in
technical aspects of Internet Communication. Groups such as the University of Padua,
Trento and Edinburgh are more involved in ontology and knowledge representations and
are discovering the importance of identity in defining domain specific ontologies
(Guarino & Welty 2000; Guarino et al. 1994).
The research approach will be multidisciplinary, touching areas such as Artificial
Intelligence, Information Sciences, Cognitive Sciences and Philosophy. The modelling of
agent internal processes will take a defined approach with internalisation of conceptual
knowledge and ontology as main areas of research. Intelligent Agent internalization of
identified search routes together with re-identification of familiar ones can expedite, to a
great extent, massive internet searches by eliminating duplication of responses in multiple
agent searches; it also eliminates cognitive overload in computer users when examining
searches results. Intelligent Systems express subjective traits in communicating with
humans and will be able to use communicative tools, cater for introspective activity
capable of learning and internalise external environment data and subjectively externalise
internal knowledge.
The design of an intelligent agent architecture will provide conceptual awareness of the
agent self and use internalisation/externalisation software agent properties in the areas of
internet e-business and Business Intelligence related activities. In eCommerce
environments a collaborative agent with introspection will be invaluable in particular in
the areas of eProcurements and eAuctions where the agent can identify and compare his
internal biddings with external agents bidding activities. On-line bidding is known by
several names, including ‘electronic reverse bid auctions’, ‘reverse auctions’ or simply ‘eAuctions’. Software agents can be used to perform automatic e-Auctions. Factors other
than price (e.g. delivery, quality, etc.) should be taken into account by the Agent prior to
the on-line bidding in order to ‘weigh’ the price bids in which to determine the overall
position of the agent bidder. There are many ways of setting up an e-Auction software
agent which will affect the way in which a bidder will be able to view his own ranking.
For instance, an agent bidder may see his position on a graph amongst other bidders’ bids
or he may see his own bidding data captured and stored inside the agent for internal
processing (Velan 2009) Identification and re-identification of eCommerce transactions
are also important for the analysis and monitoring of good transaction processes with the
related control of non conforming corporate transactions and Internet frauds. In the area
of Law enforcement and Defence intelligence, the identification and re-identification of
intelligence reports, signals or message patterns and the relevant knowledge
internalization makes the proposed agent under research an ideal internet 007. In the area
of Business Intelligence the capability for multiple intelligent agents to capture, collate
and internalize data and subsequently organise subjective knowledge using inductive
inferences is necessary in order to build knowledge bases related to external events. The
research will investigate also the analysis of Intelligent Systems internal inferential
networks of conceptualized identification with the aim of using network and lattice
theories for performing a space-time conceptual analysis of agent internalized knowledge.
1.7 Theoretical Stance
From the computer science perspective, agents are considered as autonomous and
asynchronous distributed processes with their distinct objective related traits. From the AI
(Artificial Intelligence) perspective agents are considered communicative, intelligent and
rational with the possibility of intentional communication so that they could qualify for
subjective traits. These two perspectives require a different modelling approach. While the
first perspective has the object-oriented characteristics of software tools, the second implies
intelligent communication and subjective identity and therefore requires a subject-oriented
paradigm. While object-oriented modelling has been well researched in the last decade
subject-oriented models are still in an early research phase and require a multidisciplinary
approach. By using a mediated activity model that includes agents’ subjective as well as objective
characteristics, the integration of computer science perspectives and AI perspectives can
be achieved. The mediated activity model is a central feature of Vygotsky’s Activity
Theory. Mediated activity involves social interaction mediated by the use of tools. An
important tool in social interaction mediated activity is language.
In comparing activity theory with cognitive science, we can argue that activity theory is
above all a social theory of consciousness and therefore activity theory models involve
the definition consciousness, that is, all the mental functioning including remembering,
deciding, classifying, generalising, abstracting and so forth, as a product of social
interactions activities mediated by the use of tools (Arvola 2010; Kuutti 2009).
This model can describe subjective intentional agent mediated activities and at the same
time will take into account the agent objectivity while communicating with other agents.
An agent is also a substitute for a range of human activities in a situated commercial or
industrial context and mediated activity is ideal for modelling agent activities where
agents’ individual actions and activities are analysed in a contextual environment. Agents’
activities are also dynamic and under continuous development in an historical time related
environment (Kaptelinin 1992; Kuutti 1991, 1996). The concept of tool mediated agent
communication activities is important for modelling multiple agent systems time related
interactions. Although agents are software programs themselves, they are conceived in
the model under investigation as virtual entities with subjective and objective qualities
operating in a contextual environment where activities are related with time and possess
historical properties.
A number of other cognitive models could be used to describe agent’s activities in a
contextual historical environment although these will miss their historical properties. A
situated action model for example could explain the contingent nature of agent activities
in a given situation. Situated action models emphasize the emergent, contingent nature of
human activity, the way activity grows directly out of the particularities of a given
situation. The focus of study is situated activity or practice, as opposed to the study of the
formal or cognitive properties of artefacts, or structured social relations, or enduring
cultural knowledge and values (Hutchins 2010). But this model focuses on a situated
activity without taking in account the organisational structure of multiple agent relations,
the accumulated knowledge or the agent historical pattern of activities internalised in the
agent knowledge base. Situated action modelling centres the analysis in a predetermined
specific environmental situation as a snapshot without taking into account historical time
properties involved in a flowing activity.
Practitioners of distributed cognition refer more to system functionality providing an
analysis removed from agent individual activity and focused on agents system’s structure.
Mediated Activity (MA) provides an asymmetric account for agent communicative tools.
The asymmetric differentiation between agent and tool in a given activity retains the
agent’s subjectivity qualities while distributed cognition and Systems Theories view agents
and software processes only as objective entities. Considering tools and agents in a
systemic approach such as distributed cognition also leads to the issue of describing both
agents and tools as similar entities because they are deemed as equal objects that are
communicating in symmetrical fashion (Le Dantec & Do 2009, Le Dantec 2010).
Considering tools and agents in a systemic approach such as distributed cognition also
leads to the issue of describing both agents and tools as similar entities because they are
deemed as equal objects communicating symmetrically.
A tool mediated activity model will analyse agent subjectivity, its internal and external
software tool processes and its objectivity in an asymmetric fashion. Mediated Activity
models provide a framework for the analysis of agent activities in a historical time frame
leading to an understanding of temporal changes in agent knowledge and learning
behaviour. This type of analysis is broadened to cover patterns of activity rather than
episodic staged situations. The asymmetry between communicative tools and agents will
provide a model where agent activity frames can be examined singularly or in an historical
succession. The range of agent activities will necessarily involve agents’ motives that are
mainly of environmental origin. While agent actions are always goal driven, agent
operations involve formal interaction of the agents’ subjectivity with agent objectivity
with the mediation of communicative tools (Gobbin 1998b, 1998c; Kaptelinin 1996;
Kuutti 1996).
The agent architecture model currently under research will differentiate between
processes at various categories and levels taking into consideration the objects to which
these processes are oriented. Activities are oriented to motives. Each motive is a conceptual
object that satisfies an agent need. Actions are the processes functionally and hierarchically
subordinated to activities and they are oriented at specific agent’s goals. Actions are
realised through step-by-step operations that are also determined by the actual condition
of activity and its environment (Kaptelinin 1992; Kuutti 1991). The investigated model is
also a useful tool in the analysis of cooperative agent’s team activities supporting the BDI
(Belief, Desire, and Intention) agent architectures (Busetta et al. 1999). Conceptual metarepresentations and relevance theories can also be applied to agent architectures with
agent internal activities designed for cooperative communication concepts (Edmond et al.
1998; Wiederhold 1994; Doran et al. 1997; Sperber 2000).
1.8 Research Methodology
The methodology used in this research will be based on the following three disciplines:
Information Sciences: Intelligent Systems, Artificial Intelligence, Intelligent Software
Agents, Knowledge Based Systems;
Cognitive Sciences: Communication, Learning, Belief, Desire, Intention, Agent
Societies, Mediated use of Language and semantic tools, Neuroscience;
Philosophy: Applied Ontology, Identity, Objectivity, Subjectivity Metaphysic, Inferential
Behaviour, concept of Self, Consciousness.
While this investigation falls under the Information Systems Research category, there are
disciplinary aspects related to Artificial Intelligence such as Philosophy and Cognitive
Sciences that require dissertation methods that are specific to these disciplines. The
author has been involved in research involving all three disciplines providing an
integrated approach thus focusing on the primary research objectives. This research is
mainly qualitative and in part theoretical in philosophy and cognitive sciences disciplines.
The information science discipline will use Information System research methods with
the construction of an intelligent agent meta-level architecture in order to test agent
communication theories on subjectivity collecting modelling data to support the
theoretical work.
1.9 Thesis organisation and layout
Chapter 1 Introduction:
In this chapter the thesis is introduced with research origins, background theory, issues, objectives, goals
and methods used in the research.
Chapter 2 Literature Review:
In this chapter multidisciplinary contemporary relevant literature in the field of intelligent systems
communication with key themes and unresolved issues will be presented. Literature relevant to Intelligent
Systems (IS) with Philosophy and Cognitive Sciences aspects relevant to IS will be reviewed providing an
integrated literary review focused on the research goals.
Chapter 3 - Activity Theory and Intelligent Systems Communication:
Chapter 3 will focus on subjective intelligent communication theory between intelligent systems. The use of
semantic and linguistic tools mediated activities will be discussed and a number of models presented. ToolMediated Communication Activity and Activity theory application to intelligent agents will be presented.
Foundation theories of objective internalisation and subjective externalisation of knowledge during
intelligent communication exchanges and intelligent systems communication symmetry will be discussed.
Chapter 4 - Subjective Communication Issues:
Intelligent agent subjective and objective communicative properties as described in Chapter 3 pose
philosophical issues of artificial internal conscious mechanisms with objects identity in time and space and
self identification issues. The role of artificial self, intelligent agent self identification in time and space
applied to Artificial intelligent systems will be addressed.
Chapter 5 – Subjective Intelligent Systems Ontology Applications:
In this chapter I will examine the implementation of applied ontology in subjective intelligent systems
communication with emphasis on the use of formal ontology applied to intelligent agent systems. The
important area of identity and subsumption and an investigation on the internal/external ontology
domain criteria applied to intelligent agent subjective communication exchange will be carried out.
Chapter 6 - ALMA Subjective Intelligent System Test Model:
In this chapter the design and development of novel intelligent agent meta-architectures for experimenting
subjective communication will be presented. Subjective/objective properties and new system analysis design
methods for intelligent agent applications will be discussed and an. Agent Mediated Language
Architecture (ALMA) will be presented as meta-intelligent agent architecture for the modular design of
multi platform distributed intelligent software agent operating in World Wide Web environments.
Chapter 7 - ALMA Model Subjective Communication Analysis:
In this chapter I will present the qualitative analysis of the data collected in the experiments with the
ALMA model and discuss the role of subjectivity and egocentric communication noted during the testing
Chapter 8 - Conclusions:
This chapter will present the research conclusions, research position, promising themes and plans for
further research.
The diagram below in Fig. 2 illustrates the structure of the thesis.
Chapter 2: Literature Review
2.1 Literature review introduction
Intelligent Systems research activities, focused mainly on knowledge engineering and
robotics, have advanced into the area of intelligent software agency applications creating
the basis for renewed interest in the area of intelligent software agent technology. In
Franklin and Gaesser’s 1996 seminal work “Is it an Agent or just a program? Taxonomy
for autonomous agents” an initial taxonomy of agents was proposed differentiating
software agents from artificial life agent and posing the stepping stone for further
taxonomy development (Franklin & Gaesser 1996). Further developments in Internet
and agent technology enabled Matthias Klush to enhance Franklin and Gaesser’s
taxonomy adding to the Task-Specific Agents category a new subcategory of information
agents with related non-cooperative and cooperative subcategories (Klush 2000). The
taxonomy enhancements were necessary due to enhanced agent communication
capabilities and because the new intrinsic communication models available to cooperative
information agents were opening a new horizon for further research in the intelligent
systems area. Klush’s taxonomy described the cooperative line of information agents in a
similar fashion to the non cooperative one. Klush described in the preface the challenge
of collaboration among heterogeneous and autonomous agents in an open environment
and indicated as an important emerging issue the underlying communication aspects of
agent to agent cooperation and collaboration (Klush 2000).
A number of information agents, as for example MetaCrawler (Etzioni & Weld 1995)
and SIMS/ARIADNE (Knoblock, Arens & Nsu 1994), are just search bots or single
mediators, while even non-cooperative adaptive personal assistance agents are just very
smart tools. What is different in cooperative agents is that the cooperative
communicative activity could be mediated by an agent language tool and such a tool,
must be culturally fit to address goal driven cooperation activities (Gobbin 1998b).
Cooperative activities will be defined in this research as the activity of multiple entities
(Humans, software agents, robots agents) working together to achieve work and business
goals. These activities require agents to have a higher level of communicative capacity
than single software agents in order to exchange information about their common goals,
their own identity, their current status and environmental data status. Multiple agent
cooperation will require a specifically defined model together with communicative tools
able to take into account the broad issue of transferring both environmental and self
knowledge in sophisticated intelligent cooperative activities. Intelligent Systems such as
Multi-agents can find applications in corporate, government, business and security
intelligence activities providing users with a higher level of communicative cooperation
and interaction not yet available in current software agents’ applications.
While in the late 1990s we were inclined to believe that large scale deployment of agents
based computer systems were close to reality, in the early 2000s international research
focused heavily on software agent technology. A number of knowledge engineering issues
such as agent communication language standardisation, ontology and networking
protocols and in particular research needed in the area of semantic web limited the early
implementation and operability of intelligent software agents. Intelligent agents are
critically dependent on having a very rich semantic infrastructure. The level of intelligence
out of individual agents is still experimental and requires the level of subjectivity
proposed in this thesis. Subjective communication exchanges provide a common, shared
intelligent semantic infrastructure so that individual software agents can be relatively
dumb in their implementation but appear to be quite intelligent in a communicative
operation (Hendler 2007).
Interoperability at data level and ontology sharing are vital and necessary elements
characterising cooperation among intelligent systems. If knowledge could be shared and
easily linked and distributed, then agents could share services and knowledge resolving an
important issue for agent cooperation.
Web Ontology Language (OWL) became a standard in February 2004 and has become a
key language for research efforts on the Semantic Web. A large number of open source
OWL ontologies are now widely available in research and are already implemented into
many industry applications (Hendler 2007).
This research thesis will model a novel agent mediated language architecture (ALMA)
that will add to the body of knowledge in inter-agent cooperation and communication
areas increasing the body of knowledge in the areas of Human-Agent Interaction (HAI)
that is vital for interfacing high level intellectual knowledge workers with business
intelligence systems tools that increase the productivity and dissemination of knowledge
in Government and Business environments.
2.2 Why subjective communication is an important research topic
Distributed artificial intelligence (DAI) is a subfield of Artificial Intelligence (AI)
dedicated to the development of distributed intelligent solutions for complex problems.
(Hewitt & Inman 1991) From the mid 1970s to 1990s the AI subfield has evolved and
diversified into Multiple Intelligent Agents technology establishing a rich and promising
research field that brings together many disciplines such as AI, computer sciences,
cognitive science and philosophy in a broad multidisciplinary effort. Agents considered
as computational entities are autonomous, rational and relying on problem solving.
Decision making and learning are the major enhancements in agent communicative
interaction and these are important aspects requiring attention for agent activities such as
cooperation, competition and knowledge exchange not only between agents but also
between agents and humans (Weiss 2001).
Modern knowledge workers require intelligent systems with complex problem solving
capabilities able to collate knowledge coming from increasingly diverse information
sources. The knowledge collation and preparation can be achieved by groups of
coordinated and cooperating agents communicating together using agent language
communication activities and common ontology systems. Research in this area will
enhance the field of Human - Intelligent Systems interaction improving ontology
translations between agent and agent and human and agent.
Standard engineering modelling is adequate for standard operational software agents that
are performing routine tasks. In software agent communicative exchanges this model is
inadequate as it does not take into account the subjectivity of communication exchanges
and their semantic nature. There are many engineering tools and services and even data
modelling applied to agents but this is not enough to constitute the kind of deep
integrated rich intelligent semantic infrastructure that is needed to make software agents
intelligent and subjective at the same time (Macal & North 2010).
While research in agent communication is providing results, knowledge engineering
problems are compounded by lack of real-world developments, in particular in the area
of Human-Agent Interaction. (HAI) The importance of HAI is paramount for the
coordination of multi-agent systems to achieve real-world goals and also for intelligent
cooperative activities between human knowledge workers and artificial intelligent agents.
DAI’s original long term goal was in fact to develop artificial intelligence systems where
agents will be able to interact as well as humans and to understand communicative
interactions among intelligent entities whether they are human or /and software agents
(Weiss 2001; Hendler 2007). This ambitious goal raises a large number of
multidisciplinary issues and currently has not been achieved. Research efforts in this area
are important for the key role they play in future distributed computer science
applications as information systems now form large networks across the globe,
processing huge amounts of knowledge and data resident in different geographical
locations. Increasing political and economic globalisation and environmental threats
require comprehensive human activities and interactions modelling exploring sociological
and psychological boundaries (Weiss 2001).
The research is posited on the role played by subjective communication and knowledge
transfer between intelligent systems and in particular between software agents and human
organisational environments. What makes this research important is that the analysis of
subjective communication activities can generate subjective agent communication
architectural models that can be used in a broad range of intelligent systems and HAI
applications. This has been achieved by using a multidisciplinary analysis covering
philosophical and cognitive aspects of human communication and subjectivity in order to
provide a firm ground on which new architecture can be built.
While researchers in the IT discipline use engineering modelling, they could encounter a
number of issues when analysing user models describing communication between human
and intelligent systems. On the other side, researchers in cognitive sciences, linguistics
and philosophy are quite sceptical about the possibility of having intelligent
communication between software agents and humans.
The resolution for this divergence is another interesting aspect of the research on
intelligent systems subjective communication as the possibility to answer the fundamental
question currently lingering in DAI multi-agent systems: “How internal intelligent
subjective qualities are to be incorporated into the intelligent systems communication
activity model? Human communication and learning activities use a great deal of
subjective communication and it is important for humans to be able to coordinate and
cooperate with intelligent systems expressing subjective traits therefore enhancing HAI at
an optimum level.
Researching an agent mediated communication architecture able to be used in different
application environments will require the creation of a multi-purpose model where the
underlying structure is based on sound theoretical grounds. By exploring the subject and
object dialectic together with internal and external ontologies this research offers a novel
intelligent systems architecture modelling approach that avoids the problem of an overspecialised architecture. The Agent Language Mediated Activity (ALMA) model under
investigation will be able to be implemented in a wide range of human activities
enhancing the use and applications of Intelligent Agents in scientific, business and
government applications.
2.3 Software Agents communication overview. What is known?
We have seen in this chapter introduction the taxonomic changes derived by the
distributed artificial intelligence cooperation activities of software agents. These
cooperative activities have increased as the World Wide Web network usage increased
worldwide demanding new applications able to satisfy Web users. Among the more
significant we can mention on-line business, commercial distribution and procurement
hubs, medical and professional knowledge mining and online searches where agents have
started to be implemented. In view of the new developments, Frankling and Gaesser‘s
agent taxonomy has been expanded with a new emergent category: Information Agents.
This new category was then sub-categorised with cooperative and non-cooperative
behaviours. (Frankling & Gaesser 1996; Klush 2000) This new addition arose from new
research flourishing in information agents involved on Web environments such as the
non-cooperatives MetaCrawler, (Etzioni & Weld 1996); SIMS/ARIADNE, (Knoblock,
Arens & Hsu, 1994); MIAOW/InphoSphere (Gehmeyr, Muller & Shappert, 1998), and
the cooperative type of information agents such as InfoSleuth, (Nodine, Bohrer & Ngu
1999) TSIMMIS, (Bergamaschi 1997; Garcia-Molina et al. 1995) KASBAH, (Chavez &
Maes 1995) FishMarket, (Noriega et al. 1998) and AuctionBot (Wurman, Wellman &
Walsh 1998).
Collaborative activities necessary for a cooperative category of information agents implied
a higher level of communication not only between software agents but also between
agents and human users in the case of Human to Agent Interaction. Intelligent
cooperative agents can possess the ability of cooperating on demand with other agents,
infer about other agent capabilities in a given environment (Klush 2000).
Multiple agent cooperation and collaboration implies a sophisticated inter-agent
communication activity and the use of a common language to enable the communication
Communicating Multi-agent Systems (MAS) are a computational system made of agents
using tools such as communication language in a situated environment. A MAS using an
Agent Communication Language tool behave in a situated environment and need to be
modeled and designed in virtue of the environment in which it operates.A MAS that uses
mediated tools for communication can have four sorts of interaction :
1. Communication agents speak with agents;
2. Operation agents use tools such as ACL language and ontology;
3. Composition tools used by agents link with other tools;
4. Presentation tools (e.g. language and ontology) manifest to agents.
As a result tool using MAS interaction amounts to the four categories above plus the
interaction with the environment which, depending on the required level of abstraction,
we may attribute to either individual tool used or the MAS as a whole (Omicini et al 2008;
Oliva 2009, 2010). MAS employing mediated communication languages and ontologies
are good candidate to perform subjective communication if their internal processing is
designed accordingly. This thesis is exploring an internal processing model that enables
MAS to communicate in a subjective way. The ALMA subjective model under research is
following the four interactions indicated in the taxonomy shown in fig 3 below
The communication language chosen for this research is FIPA ACL that is an
international standard endorsed by IEEE. Standard ACL language has been used for
research platform stability and experimental consistency. The Foundation for Intelligent
Physical Agents (FIPA) is an international organization that is dedicated to promoting the
industry of intelligent agents by openly developing specifications supporting
interoperability among agents and agent-based applications. This occurs through open
collaboration among its member organizations, which are companies and universities that
are active in the field of agents. FIPA makes the results of its activities available to all
interested parties and intends to contribute its results to the appropriate formal standards
bodies where appropriate.
A FIPA ACL message contains a set of one or more message parameters. Precisely which
parameters are needed for effective agent communication will vary according to the
situation; the only parameter that is mandatory in all ACL messages is the performative,
although it is expected that most ACL messages will also contain sender, receiver and
content parameters (FIPA 2002).
The model of agent communication in FIPA is based on the assumption that two agents,
who wish to converse, share a common ontology for the domain of discourse. It ensures
that the agents ascribe the same meaning to the symbols used in the message. For a given
domain, designers may decide to use ontologies that are explicit, declaratively represented
(and stored somewhere) or, alternatively, ontologies that are implicitly encoded with the
actual software implementation of the agent themselves and thus are not formally
published to an ontology service (FIPA 2002). Research activity in the area of
standardisation for agent communication language has not achieved yet a firm
international standard, polarising around three of the more important interaction
languages: KQML, (Knowledge Query Mark-up Language) XML, (Extensible Mark-up
Language) and FIPA ACL (Finin et al. 1994; FIPA Agent Communication Language, 1997, 2002; Cohen & Lavesque 1995).
We need to note that quite often agent communicative activities are misunderstood as
encoding /decoding processes related to the interpretation of complex linguistic codes.
The work of an encoding/decoding device is not inferential or creative. It is not
inferential because the symmetric relation between a message and a signal is quite
different from the asymmetric relation of premises to conclusion i.e. the meaning of a
sentence doesn’t logically follow from its encoded physical expression of a sound. Also
an encoding/decoding device is rarely creative. It would defeat its purpose if it were so. A
creative encoding/decoding that changes the process symmetry would generate
communication errors at every symmetry change.
The encoding/decoding process can only be a communication support process. Any
creative and inferential language mediated activity will imply some degree of knowledge
transfer. Applied domain ontology tools for knowledge representation and exchange are
used in agent communication where common meaningful descriptions are adopted
(Gruber 1993; Guarino, Carrara & Giarretta 1994; Guarino & Welty 2000).
Focusing on the interactive cooperative activities required by agents in the internet
environment we can note the relevance that knowledge representation has on such
activities. While non-cooperative autonomous agents may still communicate outside their
shell in information data exchange mode, cooperative agent communicative activities
have the intrinsic cooperative requirement to share environmental resources and goals.
Cooperation is in fact an integrated causal component in setting the stage for the
communication and use of coordinated knowledge exchange between entities (Gibson &
Ingold 1993).
Multiple agent collaboration processes, while exhibiting cooperation in order to achieve
operational goals, will require a higher level of intelligent subjective communication than
an autonomous agent capability. By intelligent communication I mean that a software
agent will requires an egocentric language tool mediation process in place during the
communication exchange, and by subjective I imply a degree of internal conceptual
reasoning and learning capable to sustain the negotiations, cooperation and teamwork
activities necessary to fully utilise the concept of agency in the Internet environment.
While programmers could embed their own subjective concepts into the software code of
agents displaying communicative properties, there is still great difficulty in modelling
design rules and semantics for a complex social interaction of human activities in the
increasingly sophisticated web environments. Research efforts in software agents’
communication currently focus on agent interactions but rarely do projects involve any
model or architectures involving subjectivity.
In the case of software agents’ communication, message exchanges can vary from service
oriented remote procedure calls (RPC) to more sophisticated semantic speech-act
messages. Research in agent communication language (ACL) standardisation has
increased in the last five years in virtue of increasing investigation on the high number of
cooperative multi-agents systems operating in the Web environment. The quality and
number of inter agent communication messages became important in facilitating
cooperation. In a cooperative agent activity a language tool allows the conveyance of
individual agent experience and acquired knowledge between members transferring
environmental knowledge and common goals information beyond the single agent.
Cooperative activity can also be enabled by using non verbal or sign language
communication. In practice any language tools either verbal or non verbal can facilitate
and mediate knowledge transfer between entities through communicative activities.
Concepts of cooperation and communication will be used in discussing a new
architectural model for intelligent software agent cooperation and communication using
language as a mediating tool. By using subjective models, language tools for
communicative exchange will enable software agents to interact while hiding the details of
their internalised knowledge and architectures. (Labrou, Finin & Peng 1999)
Research efforts in software agent communication are moving toward the standardisation
of speech-act type of languages for Agent Communication Language (ACL) protocols
(Castelfranchi & Werner 1994; Nagao &Takeuchi 1994; Dautenhahn & Numahoka 1998;
Nass, Tauber & Ellen 1994). While primitive information and knowledge exchange can
be achieved by using remote method invocation such as RPC and RMI to CORBA and
object request servers, there is a marked difference in using ACL for software agent
communication (Gensereth & Katchpel 1994).
ACL language can achieve a higher level of communication between software agents for
two main reasons:
1) The implementation of ACL language enables the handling of propositions, rules and
tasks oriented multiple software agent conversations;
2) ACL language messages can transfer declarative speech-act language expressions
enabling a higher level of belief, desire and intention activities instead of just procedures.
Software agent communication requires an exchange language tool richer than RPC as
agents need to specify desired states, goals, and knowledge content in an open
environment (Labrou, Finin & Peng 1999).
The US Defence Advanced Research Project Agency (DARPA) initiated the ACL
standardisation trend by establishing the Knowledge Sharing Effort, (KSE) with research
academic and industry as participants (Neches 1991; Patil 1997). The central KSE
concept was that cooperative knowledge sharing requires a) communication and b) a
common language. (Labrou, Finin & Peng 1999) By defining an agent common language
as a first layer KSE was trying to achieve the possible exchange of semantic content
among software agents creating a further upper layer as ontology with terms and
definitions (Gruber 1993).
With research still moving toward a definitive standard, the two major ACLs currently
used in the software agent development area are KQML and FIPA-ACL. Both KQML
and FIPA-ACL use a high level, message oriented, communication structured tool and
protocol that is independent of both content syntax and ontology. (Labrou, Finin & Peng
1999) Both KQML and FIPA-ACL are independent also from transport protocols,
(TCP/IP, SMTP, HOP, etc.) independent from content language (SQL, Prolog, etc.) and
independent of ontology declarations embedded in the messages. (ibid. 1999) KQML has
been mainly used as informal and partially semantic description (Cohen & Lavesque
1995; Labrou & Finin 1998).
Software agent cooperation requires a more effective semantic exchange and
interoperability. The Federation of Intelligent Physical Agents (FIPA), an organisation
including universities, major telecom companies and research labs, started to promote
communication standard specifications able to maximise interoperability and
communication between agents based systems. While FIPA is a standard organisation for
software agent’s intelligent communication, the inclusion of the term “physical” in the
organisation name has been adopted to include in the language specification physical
robot agents and agent-human communication. (Labrou, Finin & Peng 1999, FIPA 2002)
(FIPA - ACL Specifications FIPA
organisation is an open international collaborative organisation that comprises
universities and industrial entities active in the intelligent agent research field in Europe
and Far East Asia. The more active participants are NEC, NHK, Nortel, Siemens,
Alcatel, British Telecom, France, German and Italian Telecoms. FIPA ACL is based on a
speech-act theory where messages are considered actions or communicative activities
similarly to how speech-act is used in human languages and semantics
(; FIPA 2002).
The research development in ACL has been focusing solely on KQML and FIPA ACL
languages. Currently no Internet standardisation bodies have yet included ACL in their
standardisation projects. Use of Extended Mark-up Language XML, .Net and Resource
Definition Format could be used to increase integration. (Labrou, Finin & Peng 1999)
FIPA ACL organisation now been upgraded to the level of standard working committee
at the IEEE so this development promises a marked research re-orientation in the area of
ACL standardisation including research into the Internet environment communication
and interaction of intelligent agents (
2.4 Mediated communication activity
Using language tool mediation in the area of cooperative multiple agents’ communication
is novel but appropriate as an agent identity represents the dialectical difference between
external and internal communication activities. A subjective software agent using ACL
language tools is able to perform externalisation of internal communication activities. The
internal agent information represents the individual identity of the agent while the
externalised information represents the agent’s social identity. The contradiction between
the subjective internal agent identity information and the reality of the objective agent’s
externalised information represents the Hegelian dialectic difference or social identity of
the subjective communicative agent (Goldspink 2009)
While performing cooperative activities with other agents, an intelligent agent produces
and receives ACL speech-act messages. Although deprived of the richness of human
verbal communication, an embryonic form of communication is achieved. The same
internalisation and externalisation processes of speech and language can be modelled in
communicative agents’ internal processes that are also externalized and internalized in
sets of communicative behaviour using specific agent languages as a mediating tool (Finin
et al. 1994; FIPA 1997; Goldspink 2009).
Multiple agent cooperative systems require a range of communication activities. These
activities are necessarily mediated by the use of symbolic language tools using a common or
“translated” ontology. An important aspect of mediated activity is the use of language
tools as mediators in performing an activity hence the term mediated (Gobbin 1998a, 1998b;
Kaptelinin 1996).
A theoretical approach for cooperative multiple agents can make use of communicative
tools mediation and the cognitive view of tool mediation will apply to the use of language
or signs as mediating factors (Kuutti 1996; Lock & Peters 1996; Nardi 1996). The
dialectic process generated by subject-object relationship clearly influenced recent
theories on thought and language and their developmental processes (Parker 1993; Kuutti
1991). Cognitive scientists and philosophers quite often imply that speech and language
are mediating tools used in a communication activity. Wittgenstein was concerned with
the relationship between propositions and the world, and hoped that by providing an
account of this relationship all philosophical problems could be solved. In focusing to
reveal the relationship between language and the world: what can be said about it, and
what can only be shown, Wittgenstein in his “Philosophical Investigations” relates tools
in a toolbox with word generation functionality (Wittgenstein 1958; Coliva 2009).
.Kempson (Kempson 1977) contrasting Austin’s (Austin 1965) concept of speech-act
with Grice’s (Grice 1989) co-operative principle of communication is preparing the way
to the activity nature of language and subsequently Vygotsky’s language tool mediation
theories have been proposed in Human Computer Interaction software applications
design and research (Kaptelinin 1992).
Multiple agents perform in open operational environments that host a large number of
other agents sometimes belonging to different stake holders. (E.g. agent-mediated
marketplaces) In such multi agent systems environments (MAS) the global behaviour is
the by-product of a large number of agent to agent interactions. The agent language used
enables enhanced communication and coordination levels among agents with the
possibility to achieve either common objectives or just agent’s own goals (Zambonelli &
Omicini 2004).
Communication and coordination activities require a shared environment that will
determine the extent of coordination necessary to avoid resources contentions.
Coordination has in fact two possible outcomes depending on the environment on which
it is exercised. The first outcome is cooperation when the agent activities are not
antagonistic and the second is negotiation when agents are antagonistic or at least selfinterested as in case of e-commerce and on line negotiation (Hanhs & Stephens 2002).
In order to cooperate successfully, each agent must build and keep a coherent model of
other agent’s active behaviour including future interaction models. A cooperative MAS
system is not just a group of interactive agents, so a comprehensive architectural
modelling approach is required. The MAS models must certainly focus on the
environment on which the MAS operate and the mediated communication activities
implied by the use of an ACL tool enabling inter-agent communication exchanges
(Zambonelli & Omicini 2004; Omicini et al 2008; Oliva 2009, 2010)).
Modelling the environment in which agents operate requires identification of the basic
environment’s properties and resources available and also how the agents can interact
with it. (Omicini 2001) Modelling agent organisations is quite challenging (Parunak 1997;
Moses & Tennenholtz 1995; Shoham & Tennenholtz 1995) and requires specific
methodologies that reflect agents’ application characteristics, operational environment
and overall rules driving the MAS expected evolution. (Zambonelli et al. 2003; Omicini et
al 2008; Oliva 2009, 2010)
In software agent architectures that are centred on object-oriented design where
autonomy and pro-activity qualities are precluded by the object encapsulation property
itself, object activities can only be solicited by external service requests controls
(Zambonelli & Omicini 2004). Also the deep reliance on object oriented techniques in a
world of object ontologies, presents a modelling environment where all functionally
oriented objects are interdependent components (Bass et al. 2003). Objects operating in a
distributed systems posses a degree of software agent autonomy qualities, the
environmental context and the clear distinction between active and passive objects can
differentiates the approach between agents and their environments (Cabri et al. 2002;
Engster et al. 2003).
In an agent environment more emphasis is given to agent to agent communication
exchanges that are more complex than the object to object exchanges, thus requiring a
different modelling paradigm than the classic object oriented engineering paradigm
(Jennings 2001).
Summarising the advantages in adopting an agent oriented paradigm versus a classic
object oriented paradigm we can achieve:
a) A stronger encapsulation control form quality other than data and algorithms.
b) A better environmental separation by modelling environmental resources.
c) A higher quality dynamic level of communicative interaction thus reflecting the
complexity of modern distributed computing models.
Becoming a key component in today’s AI, software agent technology is now in the
forefront of intelligent system engineering development, enabling the integration of large
software systems exhibiting intelligent behaviour qualities. A novel approach in modelling
multiple agent systems engineering is required and a number of papers is moving forward
in this area (Ciancarini & Wooldridge 2001; Gervais et al. 2004; Ilarri et al 2008). Also
specifically suited architectural models and abstractions are necessary to improve the
potential of agent based computing (Zambonelli & Omicini 2004).
2.5 Current issues in software agent modelling and architectures
Current agent oriented architectures and modelling research areas can be categorised as:
Agent Modelling: Current agent modelling that can be purely reactive, (Parunak 1997)
logic agents (Van der Hoek & Wooldridge 2003) and Belief, Desire, Intention (BDI)
agents. (Kinny & Georgeff 1996) We still need novel architectures and modelling
approaches to deal with complex mediated communicative activities and subjective
properties of multiple agents’ models that are working in a situated environment where
dynamic ontological meaning translation and semantic interaction with human users is in
place (Zambonelli & Omicini 2004; Hachicha et al 2009; Beydoun et al 2009; Mouratidis
et al 2008).
MAS Architectures: The architectural design area is still heavily function-oriented. New
architectural approaches are starting to emerge and all of them imply the use of agent to
agent and agent to human communication where approaches range from social models to
organisational ones (Moses & Tennenholtz 1995; Omicini 2001; Oliva et al 2009, 2010
2008; Zambonelli et al. 2003; Beydoun et al 2009 ). MAS architectures following
biological organisational aspects are also been researched with promising results but they
reflect more the biologically structural than the communicative model (Parunak 1997;
Bonabeu et al. 1999; Zhang et al 2007; Omicini et al 2008; Oliva 2009, 2010). MAS’
functional architectures still embed traditional agents’ models where an improved
architecture could be necessary to achieve the communicative advantage provided by the
internet and the new advances in intelligent systems subjective communication (Adams
2004; Dennett 1994; Chalmers 1996; Zambonelli & Omicini 2004).
Traditional system analysis and design methodologies driving engineering software
development must be adjusted and possibly changed to match the communication
patterns and the intrinsic architectures driving MAS systems. While a number of novel
methodologies have been suggested recently, (Wooldrige et al. 2000; Zambonelli et al.
2003; Wood et al. 2001; Juan et al. 2002; Kolp et al. 2002) methodologies and models
supporting intelligent agents inter-subjective communication with intelligent systems and
humans will be necessary (Zambonelli & Omicini 2004).
Tools and software infrastructures are emerging to support the development of Multiple
Agent Systems (MAS) and among the various tools proposed to transform traditional
MAS specifications I can mention the adoption of AUML into agent system code and
infrastructures enabling design, support and the execution of distributed MAS
interactions (Artikis et al 2009; Bergenti et al. 2002; Cabri et al. 2002; Gomez-Sanz &
Pavon 2003; Noriega 1998; Ilarri et al 2009). A question emerges here on how we can
develop intelligent agents systems using an Agent Oriented Software Engineering
(AOSE) that can cater for future scenarios where human to agent interaction will be the
norm. The more the computational devices evolve the more complex and ubiquitous the
use of intelligent systems in portable equipment will become (Zambonelli & Omicini
2004; Omicini et al 2008; Oliva 2009, 2010)
As discussed above, computational devices with embedded agents will become so
entrenched in the way of life (offices, homes, streets etc.) that their architectures, models
and development complexity will become exponentially high and will require subjective
communication and interaction. Therefore the level of agent system engineering
abstraction and conceptual design efforts will grow accordingly. Using layered scales of
observation such as the nanoelectronics and molecular engineering industry (Currently
using Micro, Macro and Meso scales of observation) will help in the simplification and
definition of agent design boundaries that can be managed. (Zambonelli & Omicini 2004)
The use of multi-agents meta-level architectures introduced in this research thesis could
benefit AOSE distributed agents design where a virtual intelligent system can be designed
using clusters of agents residing in different geographical locations but working at metalevel of abstraction (Rahwan et al 2009; Hubner et al 2009).
2.6 Agents learning, cognition and consciousness
There is still an ongoing debate on the way intelligent software agents can possess
qualities usually ascribed to complex biological systems. This debate is still much stronger
in the philosophy of mind discipline questioning whether machines can have first-person
character of consciousness (Cordeschi 2010).Two camps are emerging; one camp holds
that experiential activities are accessible only from a first- person conscious perspective.
The other camp holds that all mental capacities are somehow computational and quite
broadly explainable in terms of information processing. (Dennett 1991)
The divide between these positions is centred on the assumption that a physical system
cannot develop and possess first-person properties by just using information processing
capabilities. Information processing is a system composed of objective parts and as such
should be third-person accessible to its internal functionality together with the system
logical, computational properties and its physical characteristics. The crucial point in the
debate is the human intuitive assumption that information processing “cannot” provide a
physical objective system with inaccessible subjective properties able to provide the firstperson characteristic of biological systems. When discussing intelligent agents Subjectivity
and Objectivity properties in this thesis I am using both Vygotsky tool mediated activity
theory and the Subject/0bject communicative relation aspects applicable to intelligent
agent-to-agents and human-to-agent communication. Chapter (4) will examine this topic
in detail.
2.7 Subjective communication, what is unknown on this research area
The use of subjective properties in the area of intelligent systems has been mooted in the
past ten years by a number of scholars (Gobbin 1998c; Gobbin 2004; Gobbin, Jentzsch
& Mohammadian 2004; Gobbin 2006; Guerin & Pitt 2001; Ricci, Omicini & Denti 2004).
While I hinted at the possibility of using mediated activity theory for intelligent agent
system communication since 1998, the proposals for using activity theory in the area of
MAS coordination started to emerge by classifying an agent coordination process
possibly subjective in nature. By adopting a subjective abstraction levels for cooperative
activity and an operational level for objective coordination (Omicini et al 2003; 2008) the
application of activity theory (AT) proposed could then be based on the classic HCI
version of AT (Nardi 1996).
This adoption unfortunately misses the essence of Vygotsky’s mediated activity theory
that involves communicative tool mediated internalisation and externalisation of
knowledge (Vygotsky 1978, 1986). However, the integration of objective and subjective
properties in multiple agents systems requires the agents to possess a novel agent
architectural model encapsulating the subjective/objective properties, following
Vygotsky’s internalisation and externalisation processes and the mediated use of linguistic
tools in inter-agent communication (Gobbin 2004; Omicini et al 2008).
While attempts to integrate subjective and objective in MAS agent coordination processes
have been achieved recently, we still do not have an agent architecture that will cater for
proper subjective intelligent agent communication and learning processes. (Omicini &
Ossowski 2003; Omicini & Rimassa 2004) Vygotsky’s cognitive internalisation and
externalisation theory then becomes important for the achievement of the kind of
cognitive inter-subjective transparency that is necessary to provide artificial subjectivity
(Adams 1994; Oliva et al 2009, 2010).
Inter-subjectivity requires a “consensus” about description of conscious first person
private and public events. This “conciliatory position” suggests that a rational agent
should move an internal or external belief closer to other agents in case of disagreement.
Disagreement between agents should not be taken simply in an objective stance as any
agent’s objective internalisation can be different by following a different ontology. Such
disagreements should prompt re-checking the internal subjective reasonings, once again
checking for more objective evidence. Current agent system architectures do not yet
have the necessary engineering models necessary to achieve intelligent inter-subjective
behaviour. However, using cognitive science theories such as recursive belief tests of an
artificial intelligent computer system based on processes identical as they are in natural
intelligence systems, no other criteria of intelligence or consciousness are needed (Rey
Current intelligent system architectures cannot yet be functionally equivalent to natural
consciousness where all knowledge of consciousness arises from first-person awareness
of conscious experience. In a third person description of intelligent agent consciousness,
we require an internal agent conscious state to be present. Current software agent
architectures certainly do not have such a state unless we consider architectures designed
involving a software agent conscious state and the programming efforts to build an agent
internal code acting as a pseudo conscious state (Chalmers 1996). However, intersubjectivity is still necessary to explain consciousness exchange between intelligent
entities in an exchange relation of conscious concepts mediated by physical interaction
tools. It is possible to establish an inter-subjective relationship even in absentia of an
appropriate medium. For example Aboriginal cave paintings convey the author’s
subjectivity even after 30,000 years. In this case the inter-subjective relation is deemed
diachronic while an immediate inter-subjective relation through communication is
deemed synchronic. Inter-subjective transparency and artificial subjectivity models will
require a novel approach for MAS architectures, as the current available models have
limitations in this area. This is proven by continuous efforts in the research community to
adopt approaches borrowed from HCI theories with the aim to improve MAS
coordination while a complete new model based on Vygotsky’s mediated activity theory
will break new grounds in the quest for subjective communication. (Omicini et al. 2006,
2008; McBurney et al 2009)
2.8 The knowledge gap we are trying to fill
We have seen above that the intra-subjects transparent communication and retention of
knowledge has a number of issues centred on inadequate intelligent systems architectural
design based on a traditional engineering model on one side, and an inadequate humanagent interaction on the other. In this thesis we are restoring the importance of tools
using intelligence where mediated activity plays a role in intelligent systems
communication by devices using tools like robots or abstract tools such as natural
language in the case of software agents.
The engineering model’s emphasis on designing intelligent systems that mimic certain
areas of human intelligence has been an excellent starting point (Amant & Wood 2005).
However, the usage of the Internet as a communication medium, and hardware
technological advances in data speed and storage, have boosted the possibility of creating
newer language tools using intelligent systems that communicate and learn from the
environment by objective internalisation and can modify their knowledge using internal
subjective tool sets. This research tries to establish the role of subjectivity in artificial
intelligent systems with the main goal of providing new architectures able to integrate
internal and external communication tools and knowledge by producing and using
communicative tools able to pass an artificial entity “Tool Use Proficiency Test” thus
providing meaningful subjective interfaces for artificial and human users.
Modern anthropological research has set the mediating use of tools as the fundamental
step in intelligent behaviour, in particular linguistic tools, for the internalisation and
externalisation of knowledge. While creating artificial intelligent tools such as robots and
intelligent software agents AI research has not yet examined in depth the possibility of
intelligent systems able to display autonomous tool use mediated activities in physical
interactions with their environment and communication with intelligent systems peers
and human users. Intelligent agent systems are in fact designed more as artificial tools
aiding human activities than artificial intelligent entities producing and using tools. AI
continued to follow the engineering model while fields of research such as cognitive
science have embarked with successful results in the area of human and primate tool use.
The wealth of cognitive science research in animal cognition and tool use can aid and
motivate the development of research on habilis artificial physical and software agents
integrating Artificial Intelligent and Intelligent Systems research (St. Amant & Wood
2005; ). Tool use is researched also in philosophical circles to test the hypotheses of tool
use influence in fostering intelligent behaviour and cognitive phenomena, (Preston 1998)
and a great deal of research work has been done on the evolution of tool use and human
cognition (Gibson &Ingold 1993; Mithin 1996; Sterenly 2003).
Vygotsky cognitive theories of tool mediation in enhancing cognitive internalisation are
important for intelligent systems and Intelligent Systems-Human interaction, but only
recently has a link been drawn between tool use and intelligent activities (Gobbin 2004;
Nardi 1996; Kuutti 1996). Intelligent systems displaying inter-subjective traits can only do
so by using a number of internal and external communication tools mediating the
exchange process. A new intelligent system architectural model is necessary to integrate
and extend the internal intelligent agent functionality with the externalised functionality of
the agent itself. Human problem solving is oriented to the external world and the use of
tools is reinforcing the externalised use of mind. Vygotsky posed the use of language and
signs as the major tools enabling the external boundary of mind to be expanded
(Vygotsky 1978).The external boundaries and extensions of mind, consciousness and self
are still an active research item complicated by an increased use of intelligent tools. The
functional extension of mind (Clark & Chalmers 1998) is now moving from hand use of
tools towards more sophisticated tools such as, for example, the mobile phone and future
intelligent systems aids. This thesis and related research will help in answering the
question of internal and external mind by research and modelling software and physical
agent architectures integrating internal and external intelligent functions.
2.9 Chapter Summary
In this chapter the relevance of our research effort in the area of software and physical
intelligent agents has been outlined stressing the areas of multiple agent communication
involving internalisation and externalisation of knowledge. A map of where we are and
what is known in the area of intelligent agent communication has been constructed using
intelligent systems engineering as well as learning, cognition and consciousness cognitive
and philosophy disciplines as relevant to intelligence and knowledge transfer processes.
The integration of these disciplines in the literature review was necessary for the
multidisciplinary aspects of this thesis and the feeling that a literature map for our
research journey needed a wider range of information on intelligent systems current
research. The current challenges in multiple agent systems modelling and architecture
have been discussed together with the latest agent interactive coordination theories such
as activity theory borrowed from HCI field. What is still unknown in this area has been
discussed in parallel with recent developments in the extension of cognition and
subjective consciousness providing direction for researching a novel model and
architecture able to address current intelligent system problems in knowledge transfer
and interaction with human agents. The new model under research will be useful also to
test problems of a cognitive nature and be able to provide artificial intelligence artefacts
aiding current research in cognitive science and philosophy of mind areas. The next
chapter will provide a more detailed analysis in the area of tool mediated activity theory
applied to artificial intelligent systems explaining in detail the internalisation and
externalisation processes necessary to establish intelligent communication patterns.
Internalisation and externalisation of knowledge mediated by communicative tools such
as language and signs will provide the theoretical foundation for subsequent thesis
Chapters discussing intelligent systems subjectivity, identity and ontology as foundation
stones for the new intelligent architecture under research.
Chapter 3: ActivityTheory and Intelligent Systems
3.1 Intelligence and Tool Mediated Activity.
Use of tools by primates and some member of the animal world have been explored in
detail in anthropology for a long time while the artificial intelligent use of tools by
software agents and robotics systems is expanding in Artificial Intelligence systems and
requires further investigations. (Gibson & Ingold 1993; Baber 2003) Disciplines such as
Cognitive Sciences have also produced research material in the area of cognitive use of
tools either physical tools to achieve environmental goals or cognitive tools, such as
language for intelligent communicative exchanges. (St. Amant & Wood 2005) Many
research goals and methods used in animal and primate cognition can overlap with
methods used in Intelligent Systems and in particular tools using software agents that can
model intelligent behaviour. The use of a range of tools as mediating entities for the
achievement of intelligent goals has been the base for describing intelligent behaviour and
the use of tools has been considered a precursor for a more enhanced use of cognitive
tools producing, storing and distributing intelligent behaviour.
The more important tools used by humans in communicating conceptual knowledge are
speech and writing tools and a clear understanding of the inter-functional relations and
interrelation between intelligence, thought and speech tools, either internal egocentric
speech or externalised subjective speech, is important because Artificial Intelligence and
Intelligent Systems researchers are starting to investigate the possible use of tools by
robotic systems and software agent. (ibid. 2005)
An analysis of former investigations in the area of thought and language relations will
show that cognitive theories offered from antiquity to our modern times range between
“fusion of thought with speech” on one hand and an absolute “disjunction and
segregation” on the other. (Vygotsky 1986) If we analyse verbal thinking separated from
speech we preclude the analysis of the intrinsic relation between language and thought.
Vygotsky’s holistic approach implies an integration of thought and speech and the use of
inner speech to imprint and organise intelligent systems internal content. If thought can
be considered as an activity using inner speech cognitive tools, it can then be evaluated
and modelled as an important factor in the transition from thought to external speech.
(Vygotsky 1986)
The rational and intentional conveyance of experiences and conceptual knowledge to
other intelligent systems requires a mediating tool such as speech in order to achieve the
goal of subjective communication with other intelligent entities. The means of
communication is generally intended to be the sound of the word while the real
intelligent communication enabler will require meaning i.e. class generalisation as well as
sound and signs. For Vygotsky meaning is an act of thought and at the same time
meaning is an inalienable part of a word so that it belongs to the realm of language as the
one of thought. Thus semantic analysis can model the development, functioning and the
structure of verbal thought which contain thought and speech interrelated. (ibid. 1986)
In order to intelligently communicate our own inner experience we require reference to
ontological known classes. The communicative aspect of ontology will be discussed in
Chapter (5) and as part of this research the class generalisation of word meaning will need
to investigate the two parts involved in communication activities. The first part involves
tool mediated communication theory presupposing ontology class generalisation and the
second relates to development of word meaning as an integrating part of an artificial
intelligent system internal processing construction as experiential knowledge needs to be
generalised in ontologies before it is possible to construct a symbolic inventory of all our
experiences with objects and related class relations. The symbols constructed by
experience must be ontologically associated with whole groups and delimited classes of
experience rather than single physical experiences themselves. A single physical
experience per se lodges in an individual cognitive consciousness system apparatus and is
not able as such to be externally communicated unless in symbolic form (Perlovsky
2006).Communication is therefore possible when mediated by symbolic concept transfers
that, in the case of speech and language tools, are in the form of word meaning of
generalised concepts. Intelligent forms of communication between intelligent entities,
artificial or natural, can only be achieved if conceptual designation of experiences are
generalised to classes of phenomena “known” to all participants of the communication
process considering a word meaning as a union of thought and speech and this can be
extended also to represent a union of conceptual class generalisation communicated
externally (Vygotsky 1986)
3.2 The development of concepts in artificial intelligent systems
The study of concept formation, important in cognitive science for the conceptual
development area is necessary to model symbolic communication and learning in artificial
intelligent systems. Due to experimental physical constraints, it has been extremely
difficult to observe the inner dynamic of cognitive development process in humans
although some advance in this area has been achieved. However, we are establishing with
this research, the possible modelling of an artificial intelligent system where the
architecture and construction development will enable experimental research of internal
artificial cognitive modules able to be observed and recorded. The Agent Language
Mediated Activity (ALMA) multi-agent cluster can be used for modelling communication
and internal ontology concept formation between artificial intelligent entities that could
become useful in modelling Cognitive, Artificial Intelligence and Philosophy of Mind
Traditional methods to study the development of concepts can be divided in two groups:
1) Investigation by Definition – used to investigate the already formed concepts
externalised by verbal definition. Its research methods will deal with the
externalised appearance of the concept formation symbolic finished
products. (Words and gestures) The method will overlook the dynamics
and the development of the concept formation. As in the black box
analogy, this method gives emphasis to the input/output verbal
communication without investigating the mental elaboration happening
inside the box.
2) Investigation by Abstractions – used to investigate the internal processes
leading to the intelligent system abstraction and concept formation. This
method disregards the role played by symbols and words used in the
formation of the concept and concentrates on the concept generalisation
and ontological class creation internal to the intelligent system under
research. (Vygotsky 1986)
The model under research will address both investigative groups as the process of
concept formation is developmental in nature, requiring a learning process where words
and signs are used as “tools” to direct goal-oriented artificial cognitive operations
necessary for concept creation and storage. Also the model will make use of Vygotsky’s
theories related to the necessity of internal verbal thinking as a developmental step
necessary for the creation of artificial mental concepts by using language, words and other
symbols as functional mediating “tools” indispensable for the creation, class
generalisation and storage of concepts. (Vygotsky 1986)
Vygotsky’s theories appeared in the Western research communities in 1960 and were
applied initially in HCI (Human Computer Interaction) and in particular Activity Theory
related to usability. (IT tool mediated activity) I advocated the use of Activity Theory for
intelligent agent systems in 1998 (Gobbin 1998) Only recently do we start to see research
projects using some form of tool mediated activity theories in Multiple Software Agents
coordination. (Omicini & Rimassa 2004) However, this research thesis will use Activity
Theory in a different way as we are interested in subjectivity and Vygotsky
communicative tool mediated theories can model intelligent systems communication
exchanges. The capability to subjectively internalise conceptual knowledge generated
from communicative patterns requires the implementation of a range of communicative
tools in order to transfer messages in a context of intelligent agent cooperative activities.
Cooperation, defined as the activity of multiple agents working together, requires
software agents to have communicative capacity in order to exchange information about
their common goals, their own identity, and their current status. While performing
cooperative activities Intelligent Agents can communicate by using Agents
Communication Language tools based on Speech-Act type of messages. The same
internalisation and externalisation processes of speech and language can be modelled in
communicative software agents’ internal processes that are also externalized and
internalized in sets of communicative behaviour using specific agent languages as a
mediating tool.
Intelligent systems require an architectural model with the following characteristics:
1. The ability to use symbolic patterns as communicative tools mediating agent
cooperative activities.
2. Subjective and objective qualities that are required by intelligent agents to
perform bi-directional multiple communication activities.
3. The ability to internalise representations of perceived communicative patterns.
4. The ability to externalise internally stored representations of communication
patterns to other intelligent systems.
I believe the characteristics just described are important for subjective communicative
agent interactive cooperation activities. The use of language tool mediated activity while
providing the foundation for research in human cognitive development can also provide
a theoretical framework for investigating artificial intelligent systems subjective
communication activities. (Vygotsky 1968)
3.3 Intelligent tool mediated communication activities
The asymmetry between communicative tools and agents provides a model for multiple
agents where activity frames can be examined singularly or in an historical succession.
The range of agent activities necessarily involves agents’ motives that are mainly of
environmental origin. Vygotsky’s concept of activity answers to a specific need of the
active agent: it moves toward the object of this need and terminates when it is satisfied.
Consequently, the concept of activity is necessarily connected with the concept of
motive. Activities are translated into reality through a specific or a set of actions which
are subordinated to the idea of having a conscious goal. Activities and actions are diverse
realities which do not coincide: one action can be instrumental in realizing different
activities; conversely, one motive can give rise to different goals and, accordingly, can
produce different actions. Actions are developed through operations which are
concerned with conditions. The distinction between actions and operations emerges in
the case of actions involving tools as in the case of tool mediated software agent
communication (Wang 2010)
While agent actions are always goal driven, agent operations involve formal interaction of
the agents’ subjectivity and agent objectivity with the mediation of communicative tools.
An agent can be a substitute for a range of human activities in a situated government,
commercial or industrial context and mediated activity is an ideal framework for
modelling agent activities where agents’ individual actions and activities are analysed in a
contextual environment. Agents’ activities are also dynamic and in continuous
development in an historical time related environment.
Multiple intelligent systems cooperation requires a range of communication activities.
These activities are necessarily mediated by the use of symbolic language tools using a
common or “translated” ontology. An important aspect of mediated activity is also the
use of languages as mediating tools in performing an activity, hence the term mediated. A
tool mediated activity diagram is shown in Fig. 4. (Gobbin 1998b, 1998c; Kaptelinin
1996) A theoretical approach for cooperative multiple agents can make use of
communicative tools mediation and the cognitive view of tool mediation that applies the
concept of tool mediation to the use of language or signs as mediating factors. (Kuutti
1996; Lock & Peters 1996; Nardi 1996) The dialectic process generated by subject-object
relationship clearly influenced recent theories on thought and language and their
developmental processes. (Parker 1993; Kuutti 1991)
a l fi
Cognitive scientists and philosophers often imply in a metaphorical way, that speech and
language are in fact mediating tools used in a communication activity. Wittgenstein was
concerned with the relationship between linguistic propositions and the world.
Wittgenstein in his “Philosophical Investigations” relates tools in a toolbox with word
generation functionality while Grice's work is one of the foundations of the modern
study of pragmatics. Grice studied the differences and relationships between speaker
meaning and linguistic meaning. He explained non-literal speech as the outcome of a
cooperative principle. Austin proposed a different approach associated with the concept
of the speech act and the idea that speech is itself a form of action (Wittgenstein 1958;
Austin 1965; Grice 1989).
.Vygotsky’s tool mediation theories have been recently proposed using the model
described in Fig. 4, in research on Human Computer Interaction software applications
design (Kaptelinin 1992). Vygotsky's most important contribution concerns the interrelationship of language development and thought. This concept, establishes the explicit
and profound connection between speech (both silent inner speech and oral language),
and the development of mental concepts and cognitive awareness. In this developmental
process language is considered a mediating tool.
Vygotsky’s Activity Theory has been applied to human-computer interaction describing a
"purposeful interaction" between subject and object in the world mediated by tools, both
psychological and physical. Some of these tools such as language are used in
communication to other people becoming useful for social interaction. Activity theory
can be implemented in modelling software communication.
A subjective software agent using ACL language tools is able to perform externalisation
of internal communication activities. The internal agent information represents the
individual identity of the agent while the externalised information represents the agent’s
social identity. The contradiction between the subjective internal agent identity
information and the reality of the objective agent’s externalised information represents
the Hegelian dialectic difference or social identity of the subjective communicative agent
Subjective and objective properties acquired during communication can therefore be
merged in a single agent entity as described in Fig. 5.
From a computer science perspective agents are considered autonomous, asynchronous
and using distributed processes with their distinct objective related traits. From an
intelligent systems perspective software agents are considered communicative, intelligent
and rational with the possibility of intentional communication so that they could qualify
for subjective traits. Both objective and subjective perspectives require different
architecture and modelling approaches. While the first perspective has the objective
characteristics of software tools, the second implies intelligent communication and
therefore requires a subjective paradigm. By using a mediated activity agent model with
subjective as well as objective characteristics, the integration of computer science
perspectives and intelligent systems perspectives can be achieved. The new model for
software agents under investigation can describe subjective software agent mediated
activities and at the same time takes into account the agent objectivity while
communicating with other agents.
The proposed agent tool mediated activity model analyses agent subjectivity, its internal
and external software tool processes and its objectivity in asymmetric fashion. Mediated
Activity models provide a framework for the analysis of agent activities in an historical
time frame leading to an understanding of temporal changes in an agent’s knowledge and
behaviour. This type of analysis is broadened to cover patterns of activity rather than
episodic staged situations. I consider tool mediation theories as important for the
modelling and design of artificial intelligent systems possessing a greater degree of
subjective intelligent communication capability. At this point we need to ask ourselves a
number of questions on how a different functional use of communication exchanges
might differentially determine behaviour and mastering of an intelligent system internal
subjective operations, e.g. :
How can artificial intelligent system architecture store S1 with the aid of S2? (S1 is
a physical object information, S2 is a mediating symbolic tool as for example an
agent language)
How can an artificial intelligent system internal focus be directed to S1 with the
“mediation” of S2 communication exchanges?
How can an agent language speech-act word or sign be internally associated with
S1 and externalised by using S2 tools mediation?
Theoretical aspects for each question will be discussed in Chapters (4) and (5) where an
aim, to improve the understanding on how appropriate artificial intelligent systems
architectures designs can increase knowledge in subjective developmental learning, will be
explored. This will become important in future research aimed to establish subjective
communication exchanges between human users and artificial intelligent systems such as
software agents and robots.
3.4 Development of Artificial Intelligent Systems Knowledge
Having discussed that mediated activity and the use of language tools are important
factors in the subjective externalisation of intelligent behaviour, we need now to examine
the challenging development of a system architecture which provides the internalisation
and externalisation requirements and the subsequent developmental learning necessary to
build internal knowledge. The cognitive science theories involved integrate Intelligent
Systems and artificial intelligence engineering disciplines and this subsection will try to
expose their relevance and usefulness. I start by outlining the classic Vygotsky
development of internal cognition issues that are querying the very essence of conceptual
development and the integration of language and thought mediated by tool use activity.
(Gobbin 1999) The following critical questions have been presented taking into account
the issues related to artificial intelligent systems and will be revisited during the research:
1. What is the relation between an artificial intelligent system (either software agent
or physical robot) and its environment? (In either physical sensing or
communicative exchange with other intelligent entities);
2. What new form of intelligent system activities and architectures is responsible for
the establishment of subjective communication exchanges and what is the
consequence of such activities for internalised knowledge base learning?;
3. What is the causal relationship between the active use of internal symbolic tools
(agent communication language, natural language, communication interfaces etc.
and the development of subjective intelligent communication patterns between
artificial intelligent systems?
The design of subjective intelligent systems is therefore characterised by both internal
and external mediated communicative activities that involve qualitative
transformation of one form of quantitative physical communication patterns
(objective) into qualitative subjective externalised patterns derived by internal
processing and inference. The design of intelligent system architecture must be able
to perform communicative activities using internal mediating tools able to fit in a
given communicative environment and integrate the objective communication in a
subjective fashion. This integration will be analysed in the following chapter together
with issues of objective identification.
3.5 Chapter Summary
The relationship between Vygotsky tool mediated activity and intelligent activities has
been analysed and discussed in the light of possible tool use activity by robots and
software agents. (St. Amant & Wood 2005) The development of concepts in artificial
intelligent systems environments using Vygotsky theories was proposed using a
traditional cognitive science methodology. (Vygotsky 1968) Integration of activity theory
with tool mediated communication activities was discussed together with some initial
description of subjective and objective properties necessary for mediated communication
activity. The theoretical area of subjectivity and objectivity involves theories of cognitive
identification, philosophy of the self and identity. These research areas are very important
for tools use mediated activity and intelligent systems subjective communication
processes and will be analysed and discussed in the next chapter.
Chapter 4: Subjective Communication Issues
4.1 Subjective Intelligent Communication
Artificial intelligent systems, and in particular software agents, when communicating with
human users, could create a number of challenges involving self-agency, selfrepresentation and subjective communication exchanges. Multiple agent interaction and
intelligent robots will be deployed more often in areas where computer systems interact
with humans using intelligent exchange of knowledge. Artificial cognitive processes used
in such exchanges imply an integrated use of self-representation which makes self-related
information globally available.
The capacity to engage in self-reflective cognition and to initiate voluntary self-directed
actions is an important necessary step for the analysis of an artificial representational
architecture capable of intelligent communication in ontological and evolutionary learning
terms. It seems obvious that “ontogenetic” inbuilt models and capabilities are necessary
for the emergence of an artificial subjective entity able to perform genuine cooperative
forms of mediated subjective communicative activities in a cooperative situated
The inbuilt models and capabilities should be oriented towards some form of internalised
phenomenal content able to provide communication capability in a subjective context.
The activity of detecting another intelligent entity’s subjective behaviour and decoding its
intentions or conceptual knowledge requires that the intelligent system architecture is able
to work in a subjective situated communicative environment. (Metzinger 2000)
A socially situated mediation brings forward immediately the need of internalised learned
modules capturing other intelligent entities’ conceptual properties through objective
perceptions. In this way intelligent systems can internalise the intentional and
phenomenal states of other similarly intelligent systems, artificial or not. The internalised
conceptual content created through communicative exchanges represents what Vygotsky
theorised as higher mental states. (Vygotsky 1978)
This process is highly developmental and requires an intelligent system architecture
which can provide a phenomenological ontology that can correlate subjective awareness
and allows language mediated cultural dynamic exchange from groups of subjective
intelligent systems. In order to achieve this architecture my aim is to investigate logical
relationships between artificial intelligent systems’ subjectivity, intelligent application areas
environmental descriptions and language mediated communication, with the empirical
work described in Chapters (6) and (7) of this thesis. This empirical work will try to shed
light on which AI architecture modules can provide the conditions for subjective
experience assisting the body of knowledge in the area of subjectivity.
The empirical conceptual question of subjectivity presents some difficulty to be assessed
in an artificial intelligent systems environment. However, multidisciplinary research in
neurobiological correlation of subjective consciousness can shed new light on the
functionality and working of consciousness that may be artificially experimented by
models of artificial subjectivity. (Damasio 2000)
At the moment there are two fundamental issues under research in the field of
consciousness that can be used to investigate artificial intelligent subjectivity issues. The
first issue is how an intelligent entity constructs internal mental patterns derived from
external objective sensory data. The second more important issue is how an intelligent
system, while constructing objective mental patterns, can construct a sense of subjectivity
in the act of constructing the object data. (Damasio 2000) Beside the object
representation’s construction and retrieval there is a function that is self-observing and
owning what has been constructed, as there is a subjective presence in a sort of
relationship with the objective construction. Artificial intelligent systems presenting
subjective traits must take into account the two aspects above of how objective
knowledge is created inside the system and how the system can generate the sense that
there is a subjective self that owns and observes the internal knowledge creation. An
integrated artificial intelligent system internal model and architecture must be able to
include the two primary aspects of:
1) Objective representation construction and knowledge storage;
2) The artificial system’s subjective self conscious process of observing, recording and
controlling the objective representation construction and its externalisation when
Artificial intelligent systems are built to address areas of human agency by emulating
specific functions. The rising complexity of modern human activity often requires this
kind of emulation. These systems will require a higher level of human computer
interaction and the use of natural language in communication exchanges. We can also
notice increased use of intelligent systems in communicative exchanges where the
subjective level of natural language exchange between a human user and an artificial
system will need to be expanded. This brings us to the necessity for artificial intelligent
systems to internalise, produce, modify and externalise subjective concepts. Recent
research advances in subjective consciousness research by disciplines such as
neuroscience, cognitive sciences and philosophy can be used to model artificial intelligent
systems architectures able to display some level of subjectivity. (Chalmers 2000; Damasio
2000; Metzinger 2000; Revonsuo 2000)
Before we reach the stage of presenting an architectural model for subjective exchange of
information, we need to discuss theoretical research aspects of human consciousness and
subjectivity that can be used in modelling subjective architectures. According to the
neuronal correlates of consciousness (NCC) theory, states of consciousness depend on
overall biological or physical states of an intelligent system where a neuronal state
corresponds to a correlated consciousness state. Conscious experience can be grounded
as a collection of phenomenal properties. Conscious experience relates then to a number
of phenomenal properties that are grouped in families such as vision, touch, speech, etc.
An NCC is a minimal neuronal system, N, such that there is a functional mapping from
the state of N to the state of consciousness. (Chalmers 2000) From this definition we can
state that an NCC is an intelligent system whose activity is sufficient for certain states of
consciousness allowing the possibility of:
1) Different sort of conscious states have different NCCs (e.g. vision
different from speech);
2) Different part of an intelligent system can produce a single conscious
We must take into account that different NCCs can produce different states of
consciousness that may require different methods for their investigation. By isolating
NCC systems we help the exploration of functional phenomenal states associated with
consciousness. This paradigm is valid also for research in artificial intelligent systems
architectures where modular NCC functions can be designed to correspond to a
particular cognitive agency function. This type of research is central for developing a
science of artificial consciousness being a starting point in the quest for general theories
of the relationship between physical processes and artificial conscious states. (ibid. 2000)
In the discipline of neuroscience, aspects of the process of consciousness are already
mapped to operation of specific human brain internal regions and systems. Researching
problems of the subjective self has produced good results where these regions and
systems have been located in restricted sets of neuronal areas. These discoveries, in
addition to memory and language functional areas, provide promising steps in finding a
neuronal anatomy of consciousness. (Damasio 2000) So far two types of consciousness
have been mapped:
1) Core Consciousness. A basic biological phenomenon independent from
conventional memory, reasoning and language. Provides the sense of “herenow” and is not developmental in nature.
2) Extended Consciousness. A complex biological phenomenon with many levels of
organisation evolving across the system lifetime. It also depends on conventional
memory and language communication as well. Extended Consciousness provides
the intelligent system with identity and subjectivity, sense of the self in space
dimension and individual historical time. This property will enable the subjective
temporal awareness of the past and the future along with the present that is
always available in the core consciousness. (Damasio 2000)
The two types of consciousness described above provide the description of a core self as
a transient entity, and an auto-biographical self as a more permanent entity involving
subjectivity and depending on memory storage of situations to which the intelligent
system has been exposed in a situated environment. In this case stored memory,
intelligent inferences and a higher level stored conceptual linguistic description are critical
for the generation of an auto-biographical self that can be utilised for future interaction
with the environment. (ibid. 2000) In this scheme “core consciousness” is a prerequisite
for autobiographical consciousness. The immediate sense of core self consciousness is
the first step in the knowledge process although it doesn’t permit subjective knowledge at
this stage. The extended consciousness will allow the subjective qualities to appear and to
have knowledge spanning the immediate present, the memorised past and the projected
future. It has been recently proved by neurological sciences that the “extended
consciousness” is not independent from but it is built upon the foundation of “core
consciousness”. The mechanism of the evolutionary relation and interaction between the
two types of consciousness is still under active research as intelligent subjectivity
phenomenal properties are based on their external and internal interactions with the
objective environment. The two key players in the area of consciousness are in fact the
artificial or biological intelligent organism, (Subject) the objective external world (Object)
and their relationship in the course of their historical interactions. (Damasio 2000)
Consciousness can then be described as the resulting knowledge of this interaction in the
form of two events:
a) The intelligent system is involved in a relation with some object.
b) The object in this relation is causing the artificial or biological system to change.
I need to clarify that the involved object relation and the system changes can be both
external and internal in the form of internalised and externalised knowledge changes. The
process of knowledge construction then requires an intelligent system capable of
assembling active data and information about objects external to it and the ability to
transform the data into knowledge maps to be stored. Later on, the intelligent system
internal maps can be retrieved concurrently with active external objective data forming
composite maps of both internal self and external environmental objects.
This model allows the intelligent system to experience two objective conscious forms:
1. The present objective interaction with the system.
2. The present subjective interaction with activated data maps recalled from past
occurrences where the external object has interacted with the system.
A three steps process can be assumed behind the construction of extended
consciousness. The first step constructs an account of when the intelligent system
interacts with an external object. The phenomenon is transitory and lasts a brief period of
time or for as long as the interaction exists. The second step is a gradual build up of
memories of many objective instances of past objective experiences. This second step
requires representation storage for any object interaction encountered during the system
autobiographical existence. It is important to note that each of the stored memory objects
is treated by the intelligent system as a standard object in the same way as any external
object. (ibid. 2000) It important to note that at this point we are still at the core
consciousness stage, although we can already see appearing the foundation for extended
consciousness, where simultaneously autobiographical memory map collections of the
objective external world can be revisited. The process is also matching Vygotsky theories
of higher mental state development in learning processes using mediated tools activities.
(Vygotsky 1978)
While the subject/object relationship can be explained in developmental and physical
objectivity, a unified field of consciousness can be seen as a stored memory image of an
object placed in spatial and temporal context and at the same time placed in the self
conscious context. The process through which the first mapping of a physical object
cause changes the internal mapping of an intelligent system data and contemporarily
generates a second order mapping of the subject-object relationship and subsequently an
higher order integration and enhancement of the original object mapping. (Damasio
Core subjective consciousness is the outcome of this functional structure including the
emergence of the subconscious self, knowing and an image of the environmental object
causing the subject-object relationship. Core consciousness therefore emerges from a
non-verbal second-order account of self stored maps when new object maps modify
them. (ibid. 2000) A higher step in the developmental process is necessary for complex
communication exchanges. Research in NCC of communicative verbal exchanges is
active in the area of social cognition for the reason that language implies communication
exchanges that happened in synchronisation with the system’s evolution. (Metzinger
While the core consciousness first and second order of stored mapping is a prerequisite
for its existence, cognition seems to require an integrated self representation able to
externalise self related linguistically mapped information to the external world. The
engagements in subjective actions are certainly important steps that will happen when an
intelligent system is integrated in an environment mediated by communicative tools.
Intelligent systems either artificial or biological ones must rely in cooperative forms of
social activities in order to reach the higher mental developmental forms researched by
Vygotsky. Forms of higher consciousness and in particular communication will require an
adequate social context first to start and then develop further. For this reason research on
higher forms of intelligent phenomenal content maps will require not only an NCC
reductionist analysis but also an extended analysis into subjective communication
correlates. Certain forms of phenomenal self awareness (like grasping other intelligent
system intentional aspects) require that an intelligent communicating system is situated in
a network of subjective relationships. In order to develop internal subjective
communication mapping an intelligent system expressing subjective properties must be
able to model internally the externalised subjective communicative properties of other
intelligent systems. These communicative properties are not always available in their
immediacy so they must be stored internally providing the system with its own internal
states driven by external communicative states of other systems. For this reason the NCC
core consciousness correlates require integration with higher consciousness states above
the extended consciousness described before involving social communication analysis.
(Damasio 2000) This integration requires the disciplinary relationship between neuronal,
phenomenological, sociological and cultural levels of description in researching subjective
intelligent systems.
4.2 Intelligent Agents Objective and Subjective properties integration
Objective and subjective properties can be achievable in intelligent agent system
architectures providing that an internal representation mapping is available together with
Vygotsky’s tool mediated activity and conceptual learning. New directions in explaining
subjectivity are rejecting the theory of the supremacy of the mind over the physical body
like a Cartesian “ghost in the machine” in reference to Descartes mind-body dualism.
Metzinger offers a representational approach of the phenomenal first-person perspective
derived by internal representations of the dynamical relation between subjective
representations and perceptual objects that generates representational content. The
subject/object relationship is what makes a higher phenomenal experience develop into a
subjective phenomenon bound to the internal perspectives of an individual self. As soon
as an intelligent system produces subjective communicative interactions you have a
transparent representation of episodic subject-object communicative relations. (Metzinger
The subject-object dynamic relation is similar to Vygotsky’s theoretical implementation
of higher psychological functions created by the mediation between verbal or non-verbal
signs and tools and their influence in externalised intelligent behaviour development.
While the use of tools is influencing the object of activity and is externally oriented, (a
tool leads to changes in an external object) the verbal or non verbal word or sign is a
means of internal activities aimed at the internal mapping of objects. Thus words and
signs are used as tools related to internalisation processes producing maps of external
objective operations. (Vygotsky 1978) The transparency of subject /object interaction
requires a detailed analysis of both elements comprising the interaction process: the
object and the subject. Also the subjective and objective objectivity properties analysis is
important because they apply not only to the objects and subjects in the external
communication process but also in the intelligent system internal architecture providing
the transparent identification of the self as an object in communication and interaction
with other intelligent systems as subjective objects and the objective physical objects of
the external environment.
Objectivity refers to the view that the truth of a thing is independent from the observing
subject (in our case an intelligent agent or a person). This notion entails that there are
things that exist independently from an agent or that they are external to the agent itself.
Subjectivity, on the other end, denotes that the truth of some class of externalised
statements depends on the internal states or reactions of the intelligent agent making the
statement. The notion of subjectivity is that knowledge is restricted to an agent’s own
perceptions and the object qualities experienced by the agent are subject to its
interpretations. (Mandik 1998) Many debates in the contemporary philosophy of science
and epistemology employ the notion of an Objective/Subjective distinction while in the
philosophy of the mind it is a matter of controversy whether the notion of subjectivity is
epistemic, metaphysical or both. However, in our investigation on agent subjectivity, we
omit the metaphysical question where something is objective in the case where it exists
independently from the agent and subjective otherwise. (Nagel 1986; Lycan1996)
4.3 The role of identity in intelligent exchanges
The activity of object identification and re-identification is necessary when using, for
referential identification, a single unified spatio-temporal framework in the area of
subjective intelligent and learning agents. Identification is useful also in ontology creation
to determine object properties and class assignment. Identification of the “Same” is based
on objective properties and has two meanings:
1) Objective numerical and spatial identity;
2) Objective qualitative identity.
“Same” means same objective thing and same place in relation to what is not
continuously observed. If we did not have a particular identity scheme, then a new
different spatial system for each new continuous stretch of information is needed. The
spatial system must be able to re-identify not only objective things but also places where
the object resides. Therefore identification and distinction of places are related to
identification of objects and vice versa. ((Strawson 1956; Ricoeur 1992) Individualisation
is the reverse of classification. The individualisation of an object happens only if the
object has been conceptualised first. (Strawson 1956) Also conceptualisation implies
predication. The same applies for intelligent agents; an agent must be able to
conceptualise and use predicates in order to individualise and identify. Individualisation
rests on specific designation procedures that are distinct from predication aimed at only
one specimen to the exclusion to all other members of the same class because:
1) It seems to be continuous with classification and predication;
2) It would appear to encourage language-free proper names and indicators.
You can construct such a language as Quine and other have shown, but any logical
constructions cannot be considered a language that can be spoken in concrete
communicative situations. It is an abstract language that can only be written and read.
(Pariente 1985) Quine’s description of predication is to contrast the roles of singular
terms and general terms which jointly constitute predication. Quine speaks of singular
terms as referring to objects, while general terms are true of any number of objects of
which can be predicated by those terms. Quine paves the way for the criterion of
ontological commitment by identifying its equivalents in ordinary language. (Jacquette
In identifying individual agents the strategy is to isolate among all the particulars we may
assign to a software agent, the particular property that is necessary to individualise an
individual agent. This individual privileged particular then will belong to a specific type
called “Basic Particulars”. (Strawson 1964)
Objective physical bodies and in our case objective agents will constitute such basic
particulars in the sense that nothing at all can be identified unless it ultimately refers to
those basic particulars. Therefore individuals and agents, as individuals’ proxies, belong to
a spatio-temporal schema that contains each environmental object as well as the agent
self as an object. Agents need to operate with a single and unified spatio-temporal system
of objects of which it makes sense to enquire about their spatial position. Also the own
agent spatial position in the environment needs to be determined at any one time of the
agent current activity and on the history of the activities themselves. For the reasons
above an agent should be able to re-identify a spatial position at a different time to their
activity in particular when resuming a suspended activity. The process of object reidentification in a spatio-temporal schema is important for multiple agent communicative
activities where discontinuities in identification and qualitative conceptual recurrences are
4.4 Chapter Summary
Artificial intelligent systems such as intelligent software agents and autonomous
humanoid robots are increasingly involved in a variety of agency tasks requiring intelligent
communication exchanges presenting a degree of subjectivity. Such systems need to use
and develop forms of cognitive learning and natural language capabilities through situated
environmental interaction with a range of intelligent systems of biological and artificial
nature. The architecture of such artificial systems require a new approach favouring the
appearance of a subjective core-self , conceptualised mapping of internal and external
objective events and an extended conscious level where learning and language capability
can be bootstrapped. Adopting recent neuronal research theories, I have investigated in
this chapter the theories involving the conceptual feedback transformation of subjectivity
in self conscious awareness of internal memory mapping concepts.
A critical analysis of the Neuronal Correlate of Consciousness (NCC) has been provided
using Chalmers’ philosophical explanations together with a theoretical analysis of how the
different parts of a subjective intelligent system could differentiate single conscious states
such as core consciousness and extended consciousness.
I have also discussed Metzinger’s theory of the transparency of self consciousness as a
result of an intelligent system internal model and communication exchange using higher
form of mental mapping involving language and signs. A philosophical analysis covering
objective and subjective properties has been carried out involving the role of identity and
internal identification of the self and other external entities as an important factor related
to subjective intelligent communication. Communication between intelligent systems
implies the systems have an inbuilt identification system with common conceptual
descriptors for the external world in order to share and learn cultural and social
conceptual maps.
This chapter provides an introduction to the analysis of the role of ontology in intelligent
systems subjective communication, particularly in the area of applied ontology for the
external and internal intelligent agent architectural model we are researching.
Chapter 5: Subjective Intelligent Systems
Ontologies Applications
5.1 Formal and applied ontology
Intelligent Systems subjective communication requires an understanding of the
environmental world in which the communication activity is performed and the related
semantic meaning of objects and events related to this world. A subjective intelligent
system must be able to carry out ontological categorisation processing and identification
of objects and agents involved in a specific applied domain. Human to agent
communication could involve applied ontologies related to a wide range of human
activities. Ontology has been used in intelligent systems communication and is the
foundation of software agent conceptual knowledge processing. Today the terms and
rules of classic ontology are used in diverse fields and applications when ontological
analysis is required.
However, the way the ontology term is currently used can generate problems of
definition. In classic philosophical ontology, the definition of existence is quite simple but
its implications are easily a source of misunderstanding in what exists and in what sense a
physical or abstract thing exists. Before we can address the question of what specific kind
of thing exists, we need to clearly understand the concept of existence and in practice this
requires an understanding of what it means to “be”.
To find out what it means to be we must state these existential philosophical concepts in
more familiar terms, such as for example properties and analogy, but more precisely the
analysis of being can address the problem of what it means for something to exist or
better yet, why there exists something instead of nothing and also why there is only one
logically contingent actual world. The problems of pure philosophical ontology can be
adequately answered with the conceptual resources of elementary logic. The many
questions that have baffled ontology practitioners in trying to explain the nature of being
and what it means for something to exist are intertwined with the same requirement of
standard formal symbolic logic. The principles of ontology are linked in the metaphysics
of being and in this sense rely on the foundations of logic. (Jaquette 2002) I will now
introduce the distinction between the classic ontology related to metaphysics and the
“Applied” ontology used in science and currently used quite liberally in artificial
intelligence systems. (Jaquette 2002)
Ontology as a discipline is an activity that delves into philosophical problems related to
the concepts of existence and all the facts related to it. On the other side, ontology as
domain is the result of ontology discipline investigations and is generally described as
applied scientific ontology. Applied ontology constructed around an existing domain can
be subdivided as:
1. Theoretical commitment to a proposed choice of existent entities (e.g. Terrorism
ontology, Agriculture ontology, etc.) constituted by existent entities themselves
including the actual world where these entities exist. (Extant Domain) Ontology
as a theoretical domain is therefore a description of things that exist according to
a particular theory (Which may or may not be true);
2. Applied ontology as a domain is the actual world of all existing entities identified
by a true complete applied ontology theory.
In this research on subjectivity we will make use of extant applied ontologies. Any
reference to the word ontology will be intended as applied ontology. An ontology
taxonomy that specifies the applied ontology characteristics is shown in Fig. 6
As Discipline
Most general
branch of
As Domain
No ontological commitment
Cannot be interpreted as
preferred existence domain
of specific
fields of
thought and
to preferred
Actual world of
all existent
described by a
complete true
Applied scientific ontology used in intelligent agents modelling is still based on the
conceptual analysis used in classic ontology in order to be able to recommend and
describe a preferred existential domain committing itself to the existence of a particular
choice of entities. (Jaquette 2002) I have described applied ontology as scientific because
it applies the classic definition of being to determine the ontological commitment to
other areas of knowledge, sciences, maths, etc. It is similar to the way applied
mathematics in engineering is related to pure mathematics. Applied scientific ontology is
different as it can be understood as a discipline and a domain at the same time. As a
discipline, applied scientific ontology can be considered as an enquiry method
systematically identifying the categories comprising a domain of existing entities. When
we consider applied scientific ontology as a domain we must responsibly determine the
ontology of specific areas of thought and meaning which includes theoretical ontological
theories adequately listing entities that exist and can be integrated with other ontologies
belonging to the actual world (ibid. 2002). While pure philosophical ontology precedes
applied scientific ontology, a complete and correct ontology should address both sides.
Also, pure philosophical ontology must progress towards objectivity while realising that
the existence of subjective phenomena is produced by mind. For this reason we are not
free to create applied ontology domains by simply considering objects and facts
(existence) in a subjective way. As the subjective mind is tasked to explains and categorise
what it means for something to exist in an applied ontology domain problems of identity
and predication can occur (Jacquette 2002).
A property combination is a way to describe a collection of properties under a concept
term in an ontologically neutral way for a variety of purposes in the area of applied
ontology. As explained above, logic involves the possibility of predicating properties of
objects reduced to a combination of properties. This can create problems of identity
when multiple objects are possible bearers of the same kind of properties. If logical
objects are seen as the combination of properties then the identity issue disappears since
we can identify a unique logical object with any unique combination of properties as a
derivation of the Leibniz’s Law.
This Law that is also named “The Identity of Indiscernibles” is a principle of analytic
ontology. It states that no two distinct things exactly resemble each other.” The Identity
of Indiscernibles” is of interest in ontology because it raises questions about the factors
which individuate qualitatively identical objects. Leibniz’s Law is usually formulated as
follows: if, for every property F, object x has F properties, if and only if object y has F
properties, then x is identical to y (Forrest 2008; Oderberg 2009).
The attribution of a property to a logical object is then the true or false proposition that a
property belongs to the property combination associated with the object. (Strawson 1990)
The representation of objects and properties are more independent on ontological
choice. Logic is also different to both the existence of logically possible objects as well as
logically possible properties. (Jaquette 2002) The advantage of Leibniz’s Law related
property combination approach is important because takes into account all logically
possible objects identity as well as their properties related to an applied domain where
objects can be identified by using property combinations. Whenever we use applied
scientific ontology in agent communication we are not using the classic philosophical
ontology of being. We are in fact using an annotated inventory of entities we choose as
existing in a preferred existence domain in which the software agents operate. The aim of
applied ontology in the area of software agent communication is the study of the
categories or types of things that exist or may exist in a specific applied domain. The
result of such a study is a descriptive catalogue of the types of things that are assumed to
exist in the domain of interest “D” from the subjective perspective of a person that uses
a language “L” to describe “D”. (Sowa 2001) The types of categories in an applied
ontology then represent the predicates, concepts and relations of the language “L”
whenever are used to discuss objects existent in the domain “D”. The integration of logic
to applied ontology provides a pure language that can express relationships about the
entities in the applied domain of interest. While an applied ontology could be specified by
categories or types that are defined by statements in natural language, a formal ontology is
specified by a collection of names for concepts and relations organised in types-subtypes
In logic the existential quantifier is a notation for asserting that something exists. But
logic itself has not a semantic vocabulary to describe what exists. Ontology can fill this
gap and provide a study of existence, of every type of entity (Either abstract or concrete)
that makes up the world. Ontology can supply the contents for predicate calculus and the
labels that fill content to conceptual structures. The ontological content is determined by
observation and reasoning about the entity being observed. While observation provides
knowledge of the physical world, reasoning will coordinate these observations generating
the structure of abstractions that is, metaphysics. A choice of ontological categories is the
first step from databases, knowledge bases and an object oriented system design. (ibid.
The ontological categories are important for knowledge engineering and in particular for
agent to agent communication where the sharing of information requires shared
ontological categories supporting applications across a broad spectrum of business and
manufacturing activities such as engineering, accounting and sales. Applied domain
ontologies are limited to single applications in highly specialised domains. In order to
share knowledge with other domains, applications ontology must be grounded in a more
general framework.
Philosophy will provide this framework with the related guidelines and top level
categories forming a structure that can relate to the details of the lower level
categorisation properties. While philosophers build in general their ontologies from the
top down starting with great conceptions, knowledge engineers tend to work from the
bottom-up. They start in fact from specialised applied domains where a finite number of
concepts and categories are tailored for a specific domain. However, while applied
ontologies are used in specialised domains, in order to share knowledge an ontology must
be embedded in a more general framework. Philosophy can provide that framework with
its guidelines and logic related to the top level categories that can relate to a variety of
applied domains knowledge projects.
Peirce outlined a theory of predication, involving three universal categories that he
developed in response to reading Aristotle, Kant, and Hegel, categories that Peirce
applied throughout philosophy. Peirce seems to have been driven by the influence of
Kant, whose twelve categories divide into four groups of three each. Perhaps it was the
triadic structure of the stages of thought as described by Hegel. While Kant and Hegel
build up their theories on Aristotelian categories, Peirce set the three more basic
categories addressing the categories of every ontology domain:
1. The concept of being (x Independent from everything else);
2. The concept of relation (R (x, y ));
3. Concept of mediation where the first and second category are put together.
Vygotsky’s Activity Theory can adequately integrate Peirce categories model where
subjective and objective categories can be related by a mediating tool. (Peirce, 1992)
Peirce categorisation system is fundamental for the exchange of internal and external
ontologies where categories and properties are mediated by relational tools and category
appears in the vocabulary of every domain forming the foundation for ontology matching
processes by integrating subjective agent architecture ontologies with the objective
domain ontology of the environment.
5.2 ALMA Ontology Matching Techniques
Intelligent systems applications require the ontology matching to be performed
dynamically when the whole system is operating. Ontology engineering and information
integration are two areas where ontology matching is very useful. Peirce’s category system
described above uses a meta-level approach for generating new ontological categories by
viewing entities from a different perspective such as:
Firstness (Qualities inherent to an entity itself)
Secondness (Relation on reactions between different entities)
Thirdness (Mediation that allows multiple entities to perform a relational activity)
Ontology matching technology will be used to integrate external domain ontologies
based on a legal application model and internal ontologies providing dynamic
internalisation of corresponding internal concepts created during the communication
process. Usually ontology entities requiring matching are class properties or individuals
and the matching intersections are defined as alignements expressing, with various
degrees of precision, the relation between ontologies under consideration. Ontology
matching with related alignments enables the knowledge and data belonging to diverse
ontologies to integrate. (Euzenat & Shvaiko 2007) One of the more important features of
ontology matching is the use of this technology during the construction and import of
diverse ontologies. When designing domain ontology, the necessity to ensure that
relevant knowledge resources are integrated is important at the knowledge capture stage.
(ibid. 2007)
(Omicini et al. 2003)
As described in Fig. 7, the preparation of an ALMA test environment will necessarily use
ontology matching in exchange messages encapsulating applied domain ontology classes
and objects descriptions.
Current standards for software agent messages provide an adequate protocol for
declaring both the content language type and the applied ontology used during the
communication exchange. When two or more autonomous agents meet they can
exchange messages with little chance to understand one another if the same message
language protocol and ontology is not shared. Ontology matching can improve multi
agent communication processes obtaining an alignment of the communicating agent’s
ontologies. (Euzenat & Shvaiko 2007) Agents communicate by exchanging messages
using a protocol such as FIPA and these languages determine the encapsulation of the
message enabling agents to position, identify and describe themselves in an applied
interaction context. Intelligent Systems requiring communication using a different
ontological domain should use a common ontology in order to understand their
exchanged message context. There are ways for agents to perform ontology matching;
i.e. perform the matching autonomously, taking advantage of ontology alignment libraries
or using external matching services. Once ontologies are aligned, the agent negotiating
phase can start. In this phase ontology arguments and correspondence can be exchanged.
(Laera et al. 2006) Matching agreement is reached and the resulting alignment can be set
in a program that translates the exchanged message. This model will be used in ALMA to
match the internal architecture ontology that is standardised and can be reused with the
external ontology addressing a number of external domains.
5.3 Legal Intelligent Subjective Systems Application Ontology
Legal knowledge systems are increasingly used to manage the complexity and number of
norms and legislations produced by commonwealth and state governments. On-line
eGovernment services are becoming popular and citizen use online transactions more
often in the areas of Taxation and Social Security. Current web interfaces are not
providing the level of subjective interaction necessary in dealing with arguments, rules
and norms.
To create a proper test case environment dealing with subjective reasoning, I have
integrated in the research project an OWL language legal ontology interchange module
developed in Europe by the University of Amsterdam Leibniz Centre for Law under the
Framework FP6 program IST- 027655.
Federal and state governments are increasingly active in adopting metadata tags and URI
in legislative and legal files. From early schemes like EnAct adopted in Tasmania, Canada
and USA to more recent NSW initiatives like JSMS (Justice Sector Metadata Scheme) and
AGLS run by the Federal Government. Legal informatics faces two new challenges
derived by the rapid adoption by the public, legislators and legal practitioners of the
Internet semantic Web capabilities. The first challenge is to provide an open standard
foundation for the creation and implementation of OWL legal ontologies from
standardised legal documents. The second challenge is to integrate legal OWL ontology
frameworks with intelligent agents creating an intelligent Internet application
environment displaying subjectivity and learning qualities.
I will present a review of recent research and standardisation efforts in AI and Law
discussing issues in legal arguments and legal reasoning building upon emerging XML
semantic Web standards such as as MetaLex (Boer et al. 2002, 2007) and new efforts in
OWL ontology legal frameworks like LKIF. (Boer et al. 2007) Starting from the issue of
subjectivity in Human-IS interaction, this chapter will also describe the LKIF framework
and the possible integration with the ALMA research model. Law poses difficult and
interesting problems for the AI discipline. Projects solving legal problems in Case Based
Reasoning (CBR) and legal argumentation invariably provide insights into the limitation
of existing techniques creating new approaches in AI common sense reasoning.
Legal systems are not common IT system applications requiring monotonic modus ponens
reasoning. They usually touch meta-level issues like non monotonic reasoning,
representation and learning that are central to AI research. (Rissland et al. 2003) The legal
domain will require a higher degree of standardisation and integration to address the
following issues:
1. Diverse Categories of knowledge. Such as cases, rules, procedures norms and
2. Explicit styles of justifications. In Anglo-American law the doctrine of precedent cases
govern legal reasoning while in EU the style of reasoning is centred on rules and
3. Different reasoning mode. Different types of knowledge imply different types of
reasoning. For instance reasoning with case alone, rules alone, cases and rules etc.;
4. Specialised repositories of knowledge. Large collections of cases already exist for
different hierarchies of constitutional, federal, state, local and other regulatory
Within the emerging reality of the semantic Web we can distinguish a subset, which we
can call the legal semantic Web, (Sartor et al. 2008) which is constituted by the legal contents
available in the Web, that are (or at least could be) ported to OWL RDF ontologies.
Standards are being defined for identifying legal resources, so that each legal document,
when produced by any legal authority, can be unequivocally identified as a legal document
according to well specified legal XML definitions in order to create:
1. Shared standards for legal documents;
2. Standard Legal ontologies;
3. Ways of formally representing legal norms;
The coherent implementation of these standards has so far been accomplished only to a
very limited extent. Through the realization of legal semantic Web it will be possible to
improve substantially the automatic retrieval of relevant selected and integrated legal data
and the automatic processing of legal information embedded in web documents for many
different purposes. (e. g. law enforcement, deadlines control, rules application, etc.)
(Sartor et al. 2008) The advantages of using the legal field for testing ALMA models is
that the argument model is the best suited to test subjectivity. The integration of
subjective intelligent agent research with legal semantic Web standardisation will facilitate
government internal legislation processes, the maintenance of legal document sources
and the management of legislative workflows and procedures. In the eGovernment area,
the publicity of procedures and information, the dialogue between sub-national, national,
and international institutions, and the distribution of legal information will be increased.
For these reasons a definition of appropriate common standards for legal ontologies
documents can indeed provide the vital link between the production of legislation and its
use in the community.
5.4 LKIF Legal Ontology framework
LKIF core OWL ontology was developed in Europe under the EU, FP6 program IST –
027655 as a legal Ontology core framework project building upon emerging XML based
Semantic Web technologies coordinated by the University of Amsterdam Leibniz centre
for Law. OWL has been the obvious choice for representing legal ontologies and LKIFontology had to serve many purposes including reasoning with a requirement to
guarantee that the inferences remained workable. However, the developers foresee that
the support in constructing automated legal reasoning services cannot be strictly limited
to a collection of general legal terms, but should also include special legal inference and
typical problem solving and argumentation methods. (Breuker, Valente and Winkels
Although it has become common practice to describe these re-usable reasoning
structures as ‘ontologies’, we should distinguish the standard ontology model from the
dependency structures in reasoning. The reason that ontologies and frameworks became
mixed up is the fact that OWL allows the expression of more than terminological
knowledge. The confusion may be due to the belief that because knowledge is cast in
OWL it must therefore be ontology and that is a misconception according to Breuker. In
OWL one can express a large variety of knowledge, ranging from factual information
(‘individuals’) to complex relational structures that stand as a stereo-type for the use of
knowledge. (Breuker et al. 2006)
5.5 Chapter Summary
Formal and applied ontology has been discussed in order to clarify subjective aspects of
applied ontology categorisation and identification of objects. In particular the distinction
between classic ontology related to metaphysics and applied ontology related to scientific
domains was discussed aiming to use legal domain ontology and ALMA internal ontology
matching in the research tests. The advantages from using legal ontologies for the
research test on subjectivity has been described together with explanations on the
research efforts in legal knowledge engineering. The peculiar nature of legal reasoning is
such that the Legal field poses difficult and interesting challenges in solving legal
problems in case based reasoning. Meta level non monotonic reasoning frequently
encountered in Law is similar to subjective reasoning and this chapter has covered the
advantages in using Legal ontology with intelligent agent frameworks such as ALMA for
testing subjectivity. The construction and implementation of the ALMA research
environment will be described in the next chapter explaining the software and ontology
design implemented in the subjective software agent cluster model.
Chapter 6: ALMA Subjective Intelligent System
Test Model
6.1 ALMA multi-agent architecture environment description and set-up
The thesis experiment is based on a purpose built multi-agents cluster platform
integrating Vygotsky’s conceptual development and mediated communication theories.
The multi-agent cluster subjective architecture has been named ALMA, which is an
acronym for Agent Language Mediated Activity. Specific functions such as Object,
Subject and egocentric communication have been designed integrating the process of
communicative tool use to produce novel forms of subjective conceptual behaviour.
When speech and practical objective activity converges then practical and abstract
intelligence and concept creation can occur. Egocentric subjective communication plays
an important role in attaining a goal and precedes the concept internalisation process.
Egocentric subjective communication can also facilitate effective manipulation of
objective data and can control the agent’s own behaviour acquiring the capacity of being
both the subject and object of its own behaviour. In the experiments I use Java intelligent
software agent communication technology to create a functional artificial conceptual
development test environment able to model:
1. Internalisation and formation of concepts related to external ontology domain.
2. Use of the concept in egocentric internal associations.
3. Formation and subjective externalisation of judgments and new concepts.
The conceptual development model becomes important for determining the role of
subjectivity in intelligent communication, realising that objective concepts are difficult to
process without a communication language and that subjective conceptual knowledge,
apart from instinctive concepts, may not exist beyond egocentric communication.
The central activity in concept formation is a specific use of language (In our case it will
be Agent Communication Language ACL) as a functional “tool”. The challenge for the
thesis experiment is to create and test a functional intelligent agent model that can
achieve, in artificial form, the three functional conceptual development stages shown
above. The model under research uses ACL semantic as the functional communication
tool because of the intrinsic centrality of language as a mediating tool. The
communication of object information mediated by ACL messaging tools can enable the
internalisation and construction of internal concepts.
As the architecture under research is a multi-agent cluster, I will test establishing a virtual
egocentric internal communication using agent communication language. The
establishment of subjective communication between artificial intelligent systems will
become useful for the design of intelligent user interfaces and the creation of subjective
learning systems able to interface with human using subjective patterns. Subjective
intelligent systems can be applied in a range of areas such as legal consulting, security
intelligence and e-Health, using subjective interfaces communicating with and learning
from users optimised search target.
A network of multi-agents named ALMA (Agent Language Mediated Activity) is
modelled in such a way as to form meta-agent clusters integrating subjectivity and
objectivity properties and able to perform activities using communication as tools. The
ALMA agent architecture model used in my research provides a preliminary meta-level
multi-agent architecture environment as a precursor of further artificial intelligent systems
consciousness and subjectivity research. Subjectivity and objectivity and the way they
interact in allocentric and egocentric experiential representations are the main variables
involved in any attempt to answer some of the issues of artificial consciousness.
(Chalmers, 1995 Mandik 2008)
Before outlining the experiments to be performed on subjective agent communication, I
want to describe at this point again the subjectivity and objectivity properties being
researched. Due to the influence of these two properties in agent communication and
platform modelling, a short explanation at this point will be useful. (Subjectivity and
Objectivity have been outlined in detail in Chapter (4). Philosophical traditional research
contains two overall lines of thought about the relative difficulty in explaining the notions
of subjectivity and objectivity. A line of thought that can be considered as “Cartesian”,
see subjectivity as less problematic than objectivity, as the Cartesian view is centred on
the fact that knowledge is the product of the internal mind. The issue then in the
Cartesian approach is the problem to overcome the appearance of objects and their
description in mind independently from reality. Another line of thought in contemporary
Philosophy, and in particular Philosophy of Mind, reverses the problem from Cartesian
objectivity towards a subjective explanation of the objective world independently of
objective things and dependent on the subject and its mind. (Mandik 2001, 2002)
In contemporary philosophy objective things will exist irrespective of the representational
capacity of thinking them, while subjective things depend on the representational capacity
of the subject. The idea of an object can also be considered representational and
therefore concept representations must be derived from objects. (Mandik 2008) For the
purpose of the testing experiments I use ontological conceptual representations to
represent subjective and objective items. Ontological concepts are considered as objective
or subjective depending on the reference direction and representations. Subjective
representations are egocentrically centred inside the meta-agent ALMA referring only to
“internal” ontological representations (Internal representations can be representations of
external objects or of virtual self subjective nature unifying the subject/object
experience). On the other hand the types of objective representation of external objects
that do not depend on subjective representations are ontological concepts categorised as
purely objective. (Mandik 2001)
The hardest problem so far in the philosophy of mind and artificial intelligent systems is
the experiential definition of subjective consciousness. While conscious experience
derives from an objective world there is also an elusive subjective aspect of the same
experience. Metaphysics of experience is naturally paired with the idea that perceptual
knowledge rests on this sort of conscious relational link to a perceived object. But this
idea about perceptual knowledge may be hard to reconcile with Wittgenstein mind as a
close box theory as subjects and normal perceivers are equally justified in their perceptual
beliefs. Take a standard case of perceptual knowledge. In good circumstances, I see a red
ball, and, on the basis of my experience of the ball, I judge of the ball that it is red. Since
the ball and its red colour are constituents of my experience, my judgement has what I
will call an entailing experiential ground. Although my experience is not a propositional
attitude, its existence entails that the perceptual belief that I form on its basis is true
(Kennedy 2010). Chalmers’ experiential subjective consciousness theories thrives on
internal information processes and contributed to this research that is modelling
subjective software agents able to store and externalise objective experience. We are
setting the research test parameters for a non reductive theory of consciousness relying
on communication from the subject to find indications of possible subjective experience.
(Chalmers 1995, 2007)
As a short outline for a probable theory of subjective consciousness, a number of
principles that, at this stage, may form a base for a future theory of subjective
consciousness can be described. Firstly, the principle of structural coherence where
awareness can be considered as linked structurally to objective experience. Whenever there
is a conscious experience there is some corresponding information in a cognitive system
available for subjective control or externalisation report. The principle of organisational
invariance where what is important for the emergence of subjective conscious experience
is not the peculiar physical make-up of the system, but the abstract pattern of causal
organised interaction between the components “inside” the system itself. Considering the
double-aspect principle of information, while the two preceding principles act as strong
constraints with their notions of awareness and organisation, the third principle involves
the notion of information. The hypothesis that information can have two basic aspects, a
physical aspect and a phenomenal aspect can be used as explanation of the emergence of
experience from the physical objective world. Subjective experience and awareness can be
one aspect of information (Subjective) while the other side of information is embodied in
the physical aspect. (Objective) From the above, the research experiments consider the
premise that experiential representation can be subjective if they are about subjective
things and objective if are about objects. (Chalmers 1995)
The dynamic of changing representations from objective to subjective and vice versa will
be part of the experiment to try to solve the egocentric and allocentric subjectivity and
objectivity problem considering subjective representation as egocentric agent
representation and objective representation as allocentric or other-centred representation
of object, properties and relations that are independent from the subject agent itself. In
the ALMA model, the ontology conceptual representations inside the model will be
considered egocentric if related to the subject agent “I” and Allocentric or other-centred
if related to objective domain representations. The agents will use FIPA ACL (FIPA
1997) an agent speech-act type of language, as a communication tool following
Vygotsky’s Activity Theory model where thinking and learning is mediated by language
tools. (Vigotsky 1978, 1986; Gobbin 2004)
While FIPA speech-act language is more akin to gesture commands, the major semantic
role of ontology in agent communication can be used as inter agent thought processes
subjective communication. After all, domain ontology design is a though process which
encapsulates the generation and description of attributes relevant to a certain domain as
discussed in the core legal ontology LKIF that covers the main concepts common in legal
domains starting with physical, mental, roles and abstract concepts. (Breuker et al. 2006)
When working with ontologies involving concepts we need to consider that concepts
imply the notion of deriving concept attributes and infer concepts from attributes. While
the Cartesian position in mind dualism can be controversial, nevertheless the Cartesian
conceptual inference process in reasoning is a general mode of inference for inferring
conceptual attributes from a concept. Cartesian inference can be described by the
following schema:
C (A1,…..,An ) →A1│,…..,│An
C is a concept and A1….An represent its attributes and the sign │means “or”. Hence if
the concept C has attributes (A1,....,An ) then one or all of the attributes can be inferred
from it. (Saariluoma & Nevala 2007)
Information contents in artificial mental representations are dependent in their
conceptual contents and the analysis of conceptual contents and attributes can be
performed both internally and through subjective communication. There are two views
on the ontological status of concepts. One view considers a concept as a mental
representation. A second view, according to Fregean tradition, considers the status of
concept as an abstract object based on semantic meaning. (Margolis & Laurence 2007)
This research is based on artificial concepts derived by ontologies using applied domain
concepts (legal) derived from subjective analysis. I consider the domain ontology applied
in ALMA as a preliminary cluster of abstract concepts and attributes that represent a
starting point for the experiment. By integrating Vygotsky cognitive theories centred on
the origin of concepts based on goal driven speech-act recursive communication then
changes in ontology attributes inside and outside the ALMA meta-agent cluster will be
created and analysed. The research aims to analyse a model where enhanced or changed
artificial internalised subjective concepts can be externalised providing subjectivity
attributes to the ALMA meta-agent cluster as a unique intelligent entity. The integration
of Vygotsky’s Activity Theory model to the artificial intelligence area of multi agent
communication makes use of the ALMA language ACL. The ACL speech-act type of
language operating in JADE represents the mediating tool with the goal of modifying
internal and external ontology classes and attributes. (Vigotsky 1978, 1986; Gobbin 2004)
The experiment also makes use of agent ontology classes to represent internal and /or
external domain conceptual knowledge. Ontology concept formation is analysed starting
from a set of applied ontology following up the changes until the changed concept is
transferred inside the meta-agent cluster. It is also a critical feature that JADE agent
development environment can use both speech-act types of commands as well as sending
ontology classes attributes encapsulated in message transactions where conceptual
domain knowledge can be modified by sending messages between agents. As we test
subjectivity and objectivity attributes in agents, a model able to explore message
transactions content is fundamental as, in the ALMA model, three agents will represent
internals functions inside the model that will act as an intelligent virtual meta-agent.
6.2 ALMA Model Artefact Architecture
The search for the kind of relevant domain knowledge able to fully use the ALMA
architecture has been exhaustive as ALMA can be used in a variety of domains such as ecommerce, intelligence analysis, robotics, and human services areas related to intelligent
systems communication. I have chosen an application model related to the legal discipline
with tools enabling a subjective form of legal concepts and integrating activity theory,
egocentric communication and intelligent agent communication technology.
The research model will use a multi-agent architecture in the form of a meta-agent
cluster able to test activity theory internalisation and externalisation states and monitor
artificial subjectivity in intelligent systems by using a sniffer facility that is recording
agents’ message exchanges. The Alma architectural model described in this chapter uses a
cluster of communicating agents to implement an Activity Theory based intelligent
system provided by a multi-agent communicative network using an ontology framework
specifically designed for internal cognitive subjective mediated states. In Fig. 8 below I
have compiled a structure to describe the theoretical issues inherent to the integration of
the legal ontology application with the ALMA subjective system under research.
LKIF Core Ontology
Basic concepts
AI and Law
ALMA Subjective
Agent Model
Modifications to
LKIF Ontology
Pellet Reasoner
Subjective classes
Objective classes
Activity Theory
Abstract Concepts
(Relative Places, Donnely
ALMA Classes
Legal Concepts
ALMa Model
Time (Allen 1984 theory of
Protege OWL
Top LKIF Core
Case Based
Rissland et al 2003
A. Gardner 1987
Bench-Capon &
Sartor 2001
Hoekstra, Breuker
et al. 2007
Public Policy Administration
(Tax, Centrelink, Medicare)
Winkels et al 2008
An external ontology will be produced to reflect the applied domain environment where
the intelligent system operates and in this case is an OWL legal environment applied
ontology. The multi-agent cluster ALMA acting as an artificial subjective system will
subjectively externalise objective information generated by agent communicative
exchanges. The integration of information generated by the external domain ontology is
fundamental in the process of cognition and learning and the internalisation of external
domain concepts is a fundamental aspect of the ALMA multi-agent architecture
implementing activity theory concepts.
The multi-agent cluster architecture model makes use of JADE agents able to
communicate among themselves and exchange ontology information between the applied
domain and an inner conceptual agent using an experimental ontology specifically
designed for internal cognitive states. The choice of an internal ontology to provide
subjective quality to the ALMA intelligent system enables the creation of an internal
knowledge and learning environment able to model internalisation of a variety of external
applied domains and externalisation of conceptual statements relevant to the external
The design of a multi-agent cluster test platform model involved tools able to build agent
platforms suitable for testing the model and able to monitor the communication patterns
to study subjective and egocentric communication properties useful for the analysis.
Avoiding expensive proprietary software, I have used tools belonging to the GNU licence
system, currently supported by a large number of educational research institutions around
the world. The tools involved are:
Ontology Editor Protégé version v 3.3.1 – supported by Stamford University
Medical Informatics Group. USA
Java Agent Development environment JADE v3.7.1 supported by University of
Bologna, Telecom Italy and Motorola Ltd
Java IDE Eclipse Galileo v3.5.2 (SR2)
The structure of the ALMA experiment model can be seen in Fig. 9 below and a
description of the tools used in the experiment is available in Appendices B, C and E.
LKIF Legal
Ontology Java
Protégé v 3.3.1
ALMA Multi-Agents
Platform 1
JADE v3.7
ALMA Subjectivity
Experiment Test
ALMA Multi-Agents
Platform 2
Alma Ontology Java
Eclipse Galileo
v3.5.2 (SR2)
Data collection
EJADE Plugin
Reserch analysis
LKIF Ontology OWL
The ALMA container multi-agent cluster model will act as a subjective intelligent system
entity that can internalise and externalise knowledge using a mediating language tool. The
multi-agent cluster model is described in Fig. 10.
Note that individual Containers can be nested in large numbers to form complex super
containers and intelligent multi-agents aggregates that can be widely distributed around
the network. For the purpose of testing subjectivity in agent communication a JADE test
container integrating one or more clusters of specialised agents is sufficient for the
purpose of testing subjectivity and egocentric communication.
ALMA Meta-AGENT Platform
JADE Main Container
AMS Agent
DF Agent
RMS Agent
ALMA Container
Domain Applied
Subjective Externalisation
Objective Internalisation
6.3 Legal ontology conversion to Jade Java beans ontology
The tool used for OWL legal ontology conversion to Jade javabeans and ALMA ontology
design is Protégé from Stanford University Medical Informatics Group. I have followed
Protégé tools evolution for a long time noting its wide use in the ontology research
community and the excellent support provided. As explained in the last chapter the LKIF
basic legal ontology was designed by the Leibniz Research Centre of the faculty of Law,
University of Amsterdam. I had to design and integrate in Protégé the classes relevant to
the PiersonvPost legal case that is part of the test experiments.
This ontology and all the subsequent work has been produced in OWL Protégé version
4.0 and above. The research experiment is based on communication with ontology
exchanges between software agents and the multiple agents’ research platform JADE,
that we will discuss shortly, requires the use of ontologies that are in serialised Java class
format. The last version of Protégé Netbeans generator plug-in that produced ontology
classes able to be used by Jade software agents is version 3.2.1 that will run only on
Protégé version 3.3.1. The main issue encountered was that the newest Protégé version
4.1 has a Netbeans generator that produces interfaces instead of classes while the JADE
Agents ontology support is not able to deal with interfaces at the present as it is an
ongoing research project.
Version 4.0 of the Protégé netbeangenerator plug-in could not produce code that can be
used with JADE as the generated ontology class should make reference to the default
implementation rather than to the interfaces.
Protégé version 3.3.1 has been used for a long time and was a stable version. The
download is still available in the Stanford Protégé archive and was installed without
problems. Version 3.3.1 was then starting to implement the use of OWL ontologies,
although, as will be explained in the next section, I was able to merge the LKIF OWL
ontology with the agent project classes without problems.
Description of the converted LKIF legal ontology can be found in Appendix D. This
ontology contains also elements of the legal case PiersonvPost that is used for the
subjectivity experiment and is shown in Appendix G. A picture of the Protégé version
3.3.1 interface can be seen in Fig. 11 in the next page.
The ontologies shown below in Fig. 12 and Fig.13 were then transferred to an Eclipse
Java development environment where the ALMA Agent platform was created in order to
provide a test bed for the thesis modelling experiment.
The ALMA model under investigation required two applied ontologies, one external
related to an application domain as in our case the LKIF legal domain ontology, and an
internal ontology COGITO to be used by ALMA to determine learning and
externalisation. Both ontologies are available in Appendix D.
The COGITO ontology was created in Protégé’ as part of the artefact and converted to
JavaBeans ontology in order to be able to be used by JADE Agents. Both ontologies
were translated by a JADE plug-in bean generator tool shown above in Fig. 12 above
with the generated Alma_Legal ontology classes. In fig. 13 below we can see the created
OWL ontology in Protégé format.
Objects added to the Cogito ontology before conversion from OWL to JADE:
Objects added to the LKIF legal ontology before conversion from OWL to JADE:
Action Organisation
Wild animal
The following charts in Fig. 14, 15 and 16 will show how the ALMA model internal
message structure is organised for each Objective, ALMA and Subjective agent.
The Objective agent flowchart in Fig.14 in the next page shows the internalisation
filtering function performed using the external domain ontology. The domain ontology
used is the applied ontology that ALMA multi-agent cluster is using as, for example,
Business Intelligence ontology, banking ontology or as in our case, a Legal case ontology.
The domain ontology used is the applied ontology that ALMA multi-agent cluster is
using as, for example, Business Intelligence ontology, banking ontology or as in our case,
a Legal case ontology. When the ACL message is decoded and processed by the
Objective agent this agent will forward an identical message plus the positive or negative
domain ontology relevance flag to the internal Alma agent. The Alma agent is the agent
related to the Vygotsky’s egocentric subjective communication we are experimenting
The Alma agent process described in Fig. 15 below is an important section of the ALMA
model and reflects Vygotsky’s egocentric communication and the creation of subjective
concepts. Subjective thinking can have multiple phases of egocentric communication in
order to define a subjective argumentation. Each of the egocentric communication
phases can access either the domain ontology to assess a domain concept or a “cogito”
internal ontology that only this agent accesses to build a new concept in a subjective way.
The subjective egocentric process is “internal” and will be invisible to the external
domain knowledge. The only way to share this “internal” knowledge is, as Vygotsky
proposed, externalisation of subjective concepts by communication exchanges. When
finishing their egocentric communication processes ALMA multi-agent clusters models
can communicate externally by using a Subjective agent which, as shown in the process
described in Fig. 16 below is tasked to externalise to the outside domain a specific
message that the ALMA model wants to communicate to intelligent external domain
The Subjective agent therefore concludes Vygotsky’s cognitive externalisation process
relaying an ACL message that, being the product of an egocentric process, is deemed to
have subjective properties as is determined in the experimental phase of the model.
6.4 Chapter Summary
“In this chapter a subjective intelligent system test model and the theories behind the
chosen architecture have been described. The Agent Language Mediated Activity
(ALMA) Multi-agent architecture has been created with twin ontologies, one internal to
the platform and one for the external domain. The twin ontology architecture follows
Vygotsky’s Activity Theory (AT) model able to achieve objective, subjective and
egocentric functions using multi-agents structured in a particular way to achieve the
research goal of evaluating subjectivity. The model was shown in figures 14-15 and 16
where the objective, the ALMA internal and the subjective agents are indicated. Activity
Theory has been implemented in other models in particular in the area of artefact usage
ability by MAS agents to use artefacts (Omicini 2008, 2010). The ALMA subjective model
is moving this research area forward by realising that the speech-act agent language used
is the tool or artefact that the agent cluster are using. The use of Vygotsky’s AT in the
area of artefact using agents could be insufficient to map the rich semantic of cooperative
activities among intelligent systems exchanging conceptual knowledge as they are not
taking into account subjectivity. The ALMA platform possesses a subjective architecture
that can be applied to a variety of new research fields. The ALMA design will test domain
knowledge in the area of Human Agent Interaction (HAI) where agents performing a
subjective identification and re-identification of domain objects external to the agent self
will improve the subjective introspection process and will in future an agent concept of
the self. The model will also be important in the area of Internet knowledge mining and
e-Commerce where MAS can communicate and collaborate for the achievement of
common goals while being able to subjectively identify private goals against the external
domain behaviour improving knowledge mining and business intelligence processing and
interfaces. In summary Chapter 6 described the design of the ALMA subjective multi
agent research platform describing tools such as the Java Agents Development
Environment JADE, the Stanford University’s Ontology editor Protégé and the Java
software design environment Eclipse where the modelling tests were conducted. Flow
charts of the multi-agent cluster model internal and external processes were described
and discussed. Information on the ALMA platform evaluation and research contribution
can also be found in Chapter 7 and in Chapter 8”
Chapter 7 ALMA Model Subjective Communication
7.1 ALMA Single Subjective Multi-agent Cluster Model
The qualitative data collected from the ALMA model experiments collected during the
tests indicate promising roles that subjectivity can add to the area of intelligent artificial
systems communication and learning. The analysis aim is to find specific areas to help the
determination of how subjectivity can enhance communication among artificial intelligent
systems and also between users and intelligent software application tools. I have focused
analysis on two areas:
1. Subjectivity in external communication between subjective systems.
2. Egocentrism in the internal system communication when monitored by sniffer
The first point is related to the area of subjective externalisation that could be
summarised as the way intelligent entities convey their internalised concepts. It is an
important task for a communicative subjective intelligent system to determine if an
interlocutor subjective system is really subjective. Any automatic externalisation by
software agents and applications that do not perform subjective communication are easily
picked up by a subjective system after a few exchanges. So despite externalisation having
a role in expressing subjectivity, it does not always convey subjective properties if the
interlocutor is an automatic software agent or soft-bot. The second point is related to
intelligent systems subjectivity egocentric communication research as the capability of
monitoring egocentric communication is important to further refine and optimise future
subjective models and also to aid research in cognitive disciplines.
The subjective modelling experiment was conducted using two ontologies:
Alma_Legal based on LKIF legal ontology but adapted to work with Jade agents
development environment.
Cogito ontology that has been specifically created for the Alma model egocentric
communication reflecting Vygotsky cognitive theories.
I have used the term meta-agent to describe the grouping of three agents composing the
ALMA model and being part of and performing the functions related to a whole virtual
subjective agent hence the name ‘meta-agent’. The reason behind the construction of the
meta-agent artefact described in Fig. 17 was the research goal of modelling the internal
communication patterns inside the subjective meta-agent using a sniffing tool provided in
JADE with the aim of detecting subjectivity properties in the externalisation process.
ALMA Meta- AGENT Platform
JADE Main Container
AMS Agent
DF Agent
RMS Agent
ALMA Container
Domain Applied
Subjective Externalisation
Objective Internalisation
The three JADE agents Objective, Alma and Subjective were created to construct the
ALMA meta-agent platform in order to model subjective communication exchanges and
egocentric communication that could express subjective cognition and learning. In the
preliminary architectural tests a dummy software agent DA0 was created externally to the
ALMA Meta-agent model to test the initial model functionality. As the software agent
DA0 messages were created by a human, they could not represent, in the initial tests, a
subjective intelligent entity sending a message to the artificial subjective ALMA model
and receiving a message in return that could present subjectivity traits.
7.2 ALMA Single Subjective Cluster A1 Data Analysis
The JADE agent sniffer described in Fig. 18 is a valuable tool that provides a history of
the communication messages both outside and inside the ALMA meta-agent cluster
model. Two models of complete communication exchanges are shown, one providing a
confirmation to the initial request and the second one providing a disconfirmation.
The initial communication exchange starts with a Query-if message between DA0 agent
and Objective agent where DA0 is representing an external subjective entity and
Objective agent is the point of entry of the ALMA model is described in Fig 19 below.
The Query-if action is a Semantic Language (SL) action command encapsulated in the
message that ask another agent able to understand SL to confirm whether or not a given
proposition is True. The content of this SL action is a proposition string such as “Pierson
wounded fox” that uses two objects and a predicate from the PiersonvPost legal case
described in the Alma_Legal legal domain ontology and also in the legal proceeding text.
As seen at the beginning of this chapter the Alma_Legal Ontology was translated from
Protégé to JADE Java Beans and ported to Eclipse Java development environment
representing the Applied Domain Ontology in which the ALMA meta-agent cluster
performs its communication exchanges.
In the first communication exchange, DA0 agent sends a message to ALMA cluster
model querying if Pierson wounded the Fox. On the second frame in Fig. 19, the
Objective agent then receives this FIPA encoded message and decodes the message using
the Java protocol almac set to SL. The Objective agent will check the message is relevant
to the domain ontology first and then internalises a message to the Alma agent. The
ALMA multi-agent in this research experiment uses an Objective agent functioning as a
filter agent. However, it is possible in a production environment to omit this agent and
build the filtering code in a single Alma agent itself. While this technology is experimental
and may be required by research in artificial egocentric communication, the ALMA
cluster model could become a useful architecture for testing subjectivity.
As described in Fig. 19, I produced a model that is as close as possible to Vygotsky’s
theory and I set an objective agent that deals with the domain environment inputs and a
subjective agent to perform the output.
In Fig. 20 the first egocentric communication exchange starts with the Alma agent
proposing a query-if SL message to itself investigating if Pierson wounds Fox is true. The
AlmaLegal ontology is then accessed by the Alma agent that finds the objects and
predicates related to this query to be true. At this stage all the egocentric exchanges are
“inside” the Alma agent and directed to perform an egocentric validation of the
internalised knowledge. Alma agent will now send another egocentric message informing
the ontology Cogito (Self ontology) that the Proposition Pierson wounds Fox is true and
if knowledge of this is not recorded in Cogito, it will create the new classes, objects and
predicates relevant to the novel knowledge acquisition. In Fig. 20 the Alma internal agent
concludes its internal communication exchange by an inform-ref message informing the
agent that the new information assessed is due to be communicated to the external
domain environment by a Subjective agent externalisation process.
Using Inform-ref SL message type, Alma will send a message to Subjective Agent
employing AlmaLegal ontology that is the external domain ontology. The switching
between internal ontology Cogito and the external domain ontology AlmaLegal seen in
Fig. 21 is the essence of subjectivity as the externalisation of egocentrically created
concepts must be communicated externally using a language and ontology that is
understood externally to the intelligent entity. To complete the externalisation process, a
SL message is required from Subjective agent to the intelligent agent originator of the first
query if “Pierson wounded the Fox”. An inform-ref message is therefore posted to DA0
agent with the content stating that Pierson wounded the Fox concept is true as described
in Fig. 22.
The first communication model cycle between DA0 and the ALMA intelligent entity
comprised of three software agents modelled in a Subjective communication mode is
then completed. We will now perform the same experiment where a similar query will be
posed where one the SL objects is does not exist in the external and also in the internal
ontology. In this situation the ALMA intelligent entity must “learn” the new object and
subjectively respond to the agent making the original question to receive an appropriate
negative answer.
7.3 ALMA Single Subjective Cluster A2 Data Analysis
The second experimental message model was captured by the sniffer agent right after the
first message as described in Fig. 20. The DA0 external agent sends a SL query-if message
to the ALMA platform Objective agent. The message content is set to “Tom wounded
the Fox” that, from the legal case, we know is untrue and is not in the external domain
The query-if message is described in the first part of Fig. 23 showing a DA0 to Objective
ACL message using payload SL content different from the first experimental model. The
message is relayed from Objective agent to Alma agent in order to complete an
internalisation process according to Vygotsky internalisation theory. In the next stage, as
described in Fig. 24 first image, an egocentric messaging process starts in the model
where the Alma agent query-if to “itself” if the SL message payload “Tom wounded fox”
is consistent with the external domain ontology. For that purpose AlmaLegal ontology is
used in the egocentric query that will reply still in egocentric mode that “Tom wounded
fox” is not confirmed as indicated in the second part of Fig. 26.
We should note at this point that in the egocentric message uses Cogito ontology to
internalise this information as false and produce an ACL message to Subjective agent to
answer the query-if request from DA0 pointing that the statement is false. This
externalisation process is shown in Fig. 25 below.
The confirmation process began during egocentric messaging therefore it was subjective
because it originated from ALMA using a Vygotsky’s egocentric process that created a
concept variation in its ontology and also communicated this knowledge to a Subjective
agent for externalisation to be completed. Vygotsky theory that conceptual creation and
acquisition are fundamental to intelligent entities developmental learning uses internal
mental tools, like egocentric communication, to elaborate a thinking process. This
process can be replicated in artificial intelligent systems using models similar to ALMA.
The egocentric agent Alma is in Fig. 25 part one using an “inform” SL message to inform
the Subjective agent that a message with content “Tom wounded Fox “ is false and that
needs to be externalised to the external domain. Subsequently in Fig. 25, the subjective
agent prepares a SL message for the external agent DA0 with “inform-if “protocol
notifying the external agent that the “query-if” message he posted at the beginning of the
message exchange to the ALMA intelligent system is false.
This completes the modelling analysis of two communicative exchanges tests between an
external domain agent and a subjective ALMA multi-agent cluster model that was able to
internalise, and then perform Vygotsky’s egocentric communication with itself using a
Cogito ontology that is invisible from the external domain.
In the next experiment we will model and record with JADE Sniffer agent a complete
communication exchange between two subjective ALMA clusters to model the exchange
between two artificial subjective intelligent multi-agent platforms.
7.4 ALMA Double Subjective Multi-agent Cluster Model
In the first experimental model we saw a subjective communication exchange between a
standard JADE software agent and a subjective multi-agent platform ALMA. The
exchange is subjective to a half as the standard software agent does not have subjective
properties and for this reason it does not perform any egocentric communication. The
subjective process of conceptual learning from external domain messages is only
performed by the ALMA model. A full subjective exchange between two artificial
intelligent systems will introduce an ontological shift that can be shared between the two
intelligent systems’ internal mechanism of conceptual building and learning.
Vygotsky activity theory involves social subjective exchange so that internal conceptual
creation and learning in intelligent subjects is determined through communication of
concepts using “external tools” like language and applied domain ontology. The
importance for an intelligent subject’s cognitive development of possessing internal
egocentric linguistic tools is vital for Vygotsky theory on subjectivity as the verbal
thinking faculty is based on conscious awareness of verbal preparation of externalised
concepts. When two subjective intelligent systems externalise concepts, an internal verbal
thinking must be in place hidden from each other. Only the externalised statement is
visible and the possibility of internal concepts externalised in a Vygotsky subjective mode
is possible with an experimental model where both multi-agent clusters have a definitive
subjective architecture as described in Fig. 26.
ALMA Meta-Agent 1
JADE Main Container
AMS Agent
DF Agent
RMS Agent
ALMA Cluster
ALMA 2 Cluster
JADE Main Container
AMS Agent
DF Agent
RMS Agent
ALMA Meta-Agent 2
When there is a number of artificial intelligent system communicating we have
communication exchanges involving two or many clusters of software agents with
subjective externalisations. I have prepared two multi-agents subjective cluster
architectures using communicative exchanges that can form a model for an analysis of
message exchanges involving subjectivity. The experiment B has a communication
architecture that comprises two subjective meta-agents clusters made by a model
reflecting Vygotsky’s activity theory.
The first multi-agent cluster ALMA (that can be considered as a functional meta-agent
model) is a fully contained cluster with three agents set in Vygotsky’s Activity Theory
mode and including Object, Alma and Subject JADE agents. A second multi-agent
cluster ALMA2 includes the agents Object2, Alma2 and Subject2 JADE agents.
The two JADE software agent clusters are modelled so that an internal ontology Cogito
and Cogito2 is accessible only by, respectively, the Alma and Alma2 agents. As in the
preceding experiment A, an external JADE agent DA1 is used to initiate and receive
communication with the meta-agent clusters using the AlmaLegal domain ontology. In
experiment B1 and B2, I have created two conceptual models of multi-agent subjective
communication activities as described below in Fig. 27 and Fig. 36.
The first agent communication conceptual model experiment B1 shows a query
escalation situation model. It is a common situation in legal service support when an
initial query from the external domain agent DA1 to the subjective agent cluster ALMA is
assessed and may be deferred to an “expert” agent cluster ALMA2 that will answer to
both the initial querier DA1 and the referring agent ALMA. In this experiment the cluster
ALMA will accept the query using its Cogito subjective ontology and will send an
escalation message to the expert cluster ALMA2 with its own different Cogito2 subjective
ontology for reference on the initial query. Answering then occurs by informing agent
DA1 to satisfy its query and subjective agent cluster ALMA that will update its own
subjective ontology Cogito of the query solution. ALMA will then in future be able to
respond to the query without resorting to ALMA2.
The second subjective communication model exchange B2 represents a subjective debate
involving a situation when a query from external agent DA1 is posed to two agent
subjective clusters ALMA and ALMA2 at the same time. Both clusters will return
answers to the initial DA1 query that may be different, based on their subjective internal
ontology Cogito and Cogito2. The experiments B1 and B2 message exchanges are more
numerous and complex than experiments A1 and A2 involving two Jade multi-agent
clusters and taking the total number of JADE agents involved to seven.
7.5 ALMA Double Subjective Cluster B1 Data Analysis
Introducing the first subjective modelling experiment B1, it is important to note that in
experiment A1 and A2 I used a single subjective agent cluster, while in experiment B1
and the successive B2 I will model communication between two JADE agent clusters
ALMA and ALMA2. A single external agent Da1 provides the function of external
domain agent sender and receiver using domain ontology classes related to the Legal
field. (AlmaLegal) The experimental model B1 shown in Fig. 27 indicates the
communication flow between the agent DA1 and the two ALMA clusters.
A detailed communication history analysis derived from all the message exchanges
between the subjective multi-agent clusters by using an agent sniffer tool similar to the
one used in the A1 and A2 preliminary test experiments is shown in Fig. 28.
The first communication exchange begins, with JADE agent DA1 sending a “query-if
“message to the objective agent OBJ belonging to the ALMA agent cluster as indicated in
Fig. 29.
The query message asks if Mr Sanford is the counsel for the Plaintiff Mr Pierson in the
legal case Pierson vs Post at the New York Supreme Court in 1805. As seen in
experiment A1 and A2 the OBJ agent has the internaliser role among the meta-agents so
by parsing the query-if message payload it can verify the payload as belonging to the
AlmaLegal domain and subsequently send a message to the agent ALMA to complete the
internalisation of the query inside the agent cluster.
An egocentric communication message exchange is necessary to achieve Vygotsky’s
transformation of an objective concept to a subjective one. Ideally an intelligent
subjective agent, artificial or not, must be able to egocentrically validate an internalised
concept “internally” with no external view of this event to the external world. JADE
software agents can be modelled in such a way as to have the capability of sending
messages to them so by using this quality together with an activity theory defined
architectural model, we can research models of subjective exchanges between agent
As described above in Fig .30, the egocentric query-if conceptual statement is posed by
employing ALMA’s cluster internal ontology Cogito. Using this internal ontology the
agent can decide if the information requested can be provided for externalisation or in
the absence of such information it can escalate the query to another “expert” agent. With
the hypothesis that the information is not present in Cogito an accept-proposal message
is triggered in egocentric mode to the same agent.
This message will update the self contained Cogito ontology creating a new class and
objects reflecting the fact that an answer is not possible starting an “inform” egocentric
message to ALMA and subsequently starting an externalisation process for the agent
SUBJ as described in Fig. 31.
An “inform-ref” is therefore sent to the SUBJ agent in the ALMA cluster with the task of
preparing the agent SUBJ to externalise to the ALMA2 cluster the same query concept
received by the DA1 agent. The SUBJ agent will start preparation to externalise its
internal concept that Sanford is truly the Counsel for plaintiff Pierson and will begin a
subjective communication exchange with the ALMA2 agent cluster. ALMA2 subjective
agent cluster is formed by three JADE agents organised for internalisation, egocentric
introspection and externalisation processes.
As indicated in Fig. 32 the message from agent SUBJ agent is externalised to OBJ2 that
will internalise the ALMA concept that Stanford is a Counsel for Pierson to the internal
agent ALMA2 part of the ALMA2 Cluster by sending a “query-if” message. The message
uses the domain ontology AlmaLegal and the concept is internalised by ALMA2 agent
that now begins an egocentric communication process.
This process starts by firstly the agent OBJ2 querying ALMA2 agent as described in Fig.
32 about the Stanford as Counsel for Pierson concept validity using the internal Cogito2
ontology. The possibilities here are that, a) Cogito2 ontology returns “true” about the
Counsel class containing an object Sanford, b) The class Counsel exists in Cogito2 but
the object Sanford is missing, c) No class Counsel exists about the concept Counsel in
I will use the hypothesis c) No class Counsel exists about the concept Counsel in Cogito2
to experiment an agent subjective model where a “Not understood” response is sent to
itself as in Fig. 33 because the ALMA2 Cogito2 ontology cannot verify the concept as
true. In this case ALMA2 agent returns another egocentric message “propose” enabling a
change in its own ontology with new knowledge.
This is the more important aspect of a subjective intelligent agent cognitive model as it
can model the process of subjective learning. The ability for an intelligent system of
possessing egocentric communication processes and the ability to subjectively change its
internal ontology with an ontological shift is important for applying Vygotsky’s activity
theory of cognition and developmental learning to artificial intelligent systems.
ALMA2 will send an “accept- proposal” egocentric message that will trigger a change to
Cogito2 ontology to reflect the new legal knowledge concept that Stanford is the Counsel
for Pierson. The original query sent to ALMA by DA1 and relayed to ALMA2 for
reference is now in position to be sent with an “inform-if” message to DA1 and ALMA
agents externalising ALMA2 internal knowledge.
This process is performed by sending to the agent SUBJ2 the “inform-ref” message
containing the information to externalise to the external domain agent DA1 and agents
cluster ALMA replying to the original query as shown in Fig. 34. In the model an
externalisation to ALMA informs the referring cluster agent ALMA of the request result
as well as the external query originator agent DA1.
The agent SUBJ2 completes the externalisation process using the external domain
ontology AlmaLegal by sending to external agent DA1 and ALMA cluster agent SUBJ a
“confirm” message that indeed Stanford is Pierson’s Counsel and the concept is true, as
is indicated in Fig. 35. The B1 modelling experiment shows that Vygotsky’s subjective
egocentric communication is an important component for learning and can be used in
artificial intelligent systems able to record and learn new concepts and be utilised as
expert knowledge repositories.
In this experiment we have modelled an initial query posted by an external DA1 agent
that has been received by ALMA agents cluster and relayed to ALMA2 agents cluster for
consultation. Subjective intelligent agent models and architectures can be utilised in a
variety of subjective activities such as legal expert consultation, health, training and
service management environments.
7.6 ALMA Double Subjective Cluster B2 Data Analysis
The subjective agents modelling experiment B2 uses a different communicative situation
where a query is simultaneously sent to both ALMA and ALMA2 software agent clusters.
The aim of the experiment is to model an independent subjective agent response to a
legal domain query proposed by an external domain agent DA1. The two software agent
clusters have different subjective ontologies Cogito and Cogito2 that can provide
different subjective responses indicating a level of subjectivity. The diagram in Fig. 36 and
the sniffer readings in Fig. 37 show the communication flow between agent and agent
Agent Cluster
Agent DA1
Agent Cluster
process 2
The subjective communication experiment B2 will be described step by step, addressing
both agent clusters in sequence where ALMA cluster internalisation and externalisation
will be described first. Subsequently I will describe ALMA2 cluster process and compare
the two subjective communication processes response to DA1.
A sniffer agent tool has been used to record all the agent messages both internal in the
multi-agent cluster and external with a domain dummy agent.
The first process in starting the subjective experiment B2 communication modelling is
for Jade agent DA1 to set a “query-if” message with AlmaLegal domain ontology.
The message is related to the Pierson v Post legal case described in appendix G and with
the content querying if wild animals like foxes can be considered as property or not. The
double query message reaches both agent clusters simultaneously and is delivered to the
two Agents tasked to perform the internalisation process OBJ for the cluster ALMA and
OBJ2 for the ALMA2 agent cluster as shown in Fig. 38.
I begin describing the subjective communication process involving ALMA agent’s cluster
noting here that the ALMA architecture agent tasked for starting the messages
internalisation process is the JADE agent OBJ that will validate the message using the
ALMA ontology AlmaLegal and a query-ref message. The process is described in Fig. 39
together with the next internalisation step where the ALMA agent “propose”
egocentrically to itself the concept that wild animals such as foxes are not legally
considered property. We need to note here that the “internal” ontology used in the
ALMA agent cluster egocentric communication is Cogito. Vygotsky’s egocentric query
process does happen internally on the agents cluster defining the subjective cognition
ALMA agents cluster continue the egocentric communication modelling using a
“confirm” message to confirm the conceptual content of the query initiated by agent
DA1, internalised via agents OBJ and ALMA and internally validated via egocentric
communication by ALMA itself. As shown in Fig. 40, the resulting validated message is
then transferred to the externalisation process performed by the agent SUBJ. The queried
legal concept has been confirmed internally by the ALMA agent cogito ontology internal
knowledge. The ALMA cluster has completed the validation and SUBJ agent will proceed
to prepare a message to be externalised to the original querier Agent DA1 as indicated in
Fig. 41 below.
ALMA cluster agent SUBJ subjectively externalises the validated concept to the external
domain agent DA1by using an “inform-ref” message. The externalisation process uses
the external domain ontology AlmaLegal for this communication exchange. As indicated
at the beginning of the agent modelling experiment B2, agent DA1 has sent a
simultaneous query to both Alma and ALMA2 agent clusters expecting to receive back
subjective information that validates or disproves internalised conceptual knowledge. In
the second part of the experiment model B2 description starting in Fig. 42 the second
agents’ cluster ALMA2 activity model is described by using the same Agent DA1 query
and by externalising its own subjective internal knowledge using ALMA2 agent subjective
egocentric communication process and the cogito2 ontology.
A description of the ALMA2 agents cluster communication flow notes first, that the
original “query-if” message sent by external agent DA1 is received by the agent OBJ2 at
the same time as agent OBJ of the ALMA agents’ cluster. The reason the ALMA2 cluster
description is separate is necessitated by having to describe in detail both clusters
communication flow and, in particular, the different egocentric message exchanges and
subjective outcome. OBJ2 agent, upon receiving the “query-if” message and performing a
validation according to AlmaLegal ontology, will initiate the process of internalisation
necessary to start the ALMA2 agents cluster internal processes. For this purpose OBJ2
sends a “query-ref message to internal agent ALMA2 notifying the content that wild
animals like foxes are not considered property in legal terms. Upon receiving the message
from OBJ2, the agent ALMA2 starts an egocentric process by sending a “propose”
message to the self using the internal Cogito2 ontology to receive a response from the
internal ontology about the validity of the concept received.
The agent ALMA2 egocentric message exchange continues with ALMA2 sending
another egocentric message to not confirm the concept that wild animals such as foxes
cannot be property. As shown in Fig. 43, the knowledge confirmation obtained by
egocentric exchanges will need to be externalised to release the information to the
domain querier agent DA1. By examining the egocentric process we can see that, as in
the confirmation egocentric process seen in ALMA cluster, the process happens
internally in the ALMA and ALMA2 agent clusters and that the egocentric
communication model using ACL language as tool follows Vygotsky’s theory of
subjectivity. By having different externalisation from different agent clusters we can note
the importance of proposing an intelligent system specific cluster model and architecture
that will enable a subjective type of communication. These models and architectures can
be used for the design and development of intelligent agent clusters able to model
subjective interface tools to be utilised in knowledge exchange activities by intelligent
systems users.
ALMA2 agent has sent an “inform-ref” message to SUBJ2 agent with a content that do
not confirm the conceptual knowledge being queried as shown in Fig. 44. The
externalisation process begins with the Agent SUBJ2 preparing the externalisation
message to external domain agent DA1 completing the ALMA2 agent cluster
communication flow description. In the next pages an analysis of the subjective
communication model tests description A1, A2, B1 and B2 is carried out with the aim of
identifying from the data collected so far the role that subjectivity models and architecture
can have in artificial intelligent multi-agent system communication and learning. The
proposed ALMA multi-agent architecture built according to Vygotsky’s activity and
concept development theories together with the empirical modelling descriptions also
provides input to the subjectivity analysis. This analysis is qualitative and improves the
body of knowledge in the area of subjective intelligent system communication.
7.7 Externalisation of subjective properties analysis
In analysing tests A1 and A2 we could not determine from the externalised message if the
externalisation process was subjective or if the message was sent by a subjective entity. By
employing the sniffer and monitoring the internal messages my knowledge of internalised
egocentric messages did not validate the assumption that the externalised message could
have subjective properties.
However, the A1 and A2 tests were important preliminary tests to determine the model
functionality and the internal egocentric messaging. The litmus test for subjectivity
appears to be that, if the externalised message implies that an internal egocentric process
has occurred, then the message has subjective properties. The externalisation function on
itself did not provide sufficient grounds to determine subjectivity while a message
conceptual content, when externalised by the externalisation function, could show
subjective traits.
The reasons for this are that a subjective message is the result of an egocentric intelligent
system use of communicative tools. The product of an egocentric use of communicative
tools, when externalised, shows outside the system a built artefact being linguistic or
conceptual and the use of tools is the first property of cognition, learning and subjective
intelligence. In the communication exchange between two intelligent software agents, the
message content can provide an indication of which of the software agent systems is
subjective and which is an automaton where the communication activity is predictably
automatic and devoid of subjective properties.
Egocentric communication inside the multi-agent cluster was detected in experimental
ways by looking “inside” the intelligent system for patterns of egocentric communication
while the system was operating. The analysis of what is happening “inside” intelligent
entities has always been a challenge despite which modern techniques areas of internal
activity can be mapped. For this reason I have created an artificial intelligent system
model where I could monitor all the egocentric communication patterns. In the table
shown in Fig. 45 we can see a subjectivity model matrix which summarises when
subjectivity is evident due to: a) complex communication patterns, b) egocentric
communication and c) externalised message subjective content.
Not clear
Not clear
In the tests series B1 and B2 the nature of communication activity was more complex
and the egocentric role was enhanced. Two different communicative activity models were
run, one related to a reference escalation and the other to a consultation for a different
opinion in legal statements. The two activities required an increase in egocentric
communication activities and the message externalisation reflected a marked egocentric
communication exchange.
Despite the limitation posed in the tests by the ALMA software agent model because the
internal software agents were just representing subjective functionality, the results were
promising because the role of subjectivity in improving clusters communication with
subjective models agent peers was highlighted. The role of subjectivity was found also
important in the learning process area. A subjective software agent modelled like ALMA
can be utilised in learning functional clusters agents coordinating a number of intelligent
software agents sending environmental data to an ALMA cluster providing agents
coordination and leadership by learning specific agent characteristics.
7.8 Chapter Summary
The ALMA subjective software agent cluster model has been designed and
experimentally tested with different communication patterns and situations. An
application ontology related to a legal case was converted from OWL language to JADE
java beans ontology. Data from the agent clusters communication exchange was collected
and analysed for areas of subjective properties. The litmus test for finding subjectivity
appears to be that, if the externalised message implies that an internal egocentric process
has occurred then the message presents subjective properties. In order to model
intelligent agent subjective properties, intelligent software agent architectures must
include language mediated communicative activities.
A subjective message is the result of an egocentric intelligent system use of
communicative tools. The product of an egocentric use of communicative tools, when
externalised, shows outside the system a built artefact being linguistic or conceptual. The
use of tools is the first property of cognition, learning and subjective intelligence. In the
communication exchange between two intelligent software agents, the message content
can provide an indication of which of the software agent systems is subjective and which
is an automaton where the communication activity is predictably automatic and devoid of
subjective properties.
During the experiment the externalisation function on itself did not provide sufficient
grounds to determine subjectivity while a message conceptual content, when externalised
by the externalisation function, could show subjective traits. The reason for this is that a
subjective message is the result of an egocentric or internalised intelligent system use of
communicative language tools to produce an artefact that is geared to express internal
and subjective information.
Egocentric communication inside the multi-agent cluster was detected in experimental
ways by looking “inside” the intelligent system for patterns of egocentric communication
while the system was operating. The analysis of what is happening “inside” intelligent
entities has always been a challenge despite which modern techniques areas of internal
activity can be mapped.
The ALMA model was able to model the “internal areas” of agent communicative
activities determining the role that subjectivity can have in the areas of intelligent agent
communication and learning. This includes the necessity of creating an internal domain
ontology that can assist in the capture, internalisation, storage and externalization of
subjective conceptual knowledge.
This research contributes to improve the design and modelling process of intelligent
agent systems able to internalise and subjectively externalise domain knowledge
improving human-intelligent systems interaction and communication. This has been
achieved by modelling with the experiment the symmetric and asymmetric
communication exchanges between intelligent systems and by defining a new
experimental intelligent system model addressing the problem related to seeing the
internal communication process inside an artificial subjective entity.
Current multi-agent system models including recent models utilising tools and artefacts
are not able to achieve subjective status unless an attempt to model the “internal areas” of
an agent communicative activities is performed. The questions posed in researching
artificial intelligent systems communication are similar to those arising when researching
human communication. Intelligent communication activities are not mere translation
where, from a given input, an output is required
Quite often software agent communicative activities are misunderstood as
encoding/decoding processes related to the interpretation of complex linguistic codes.
The work of an encoding/decoding device is not certainly inferential or creative.
The ALMA platform experiment explored and determined the role that subjectivity can
have in the areas of intelligent agent communication and learning by evaluating all the
agent internal communication exchanges.
The analysis confirmed the exploration value of modelling and creating software agents
clusters with subjective communication capabilities and with internalised egocentric
communication properties with the ability to internalise representations of ontology
knowledge and subsequently to externalise internally stored knowledge representations to
other agents.
The analysis demonstrated that subjective qualities in artificial software agents’
communication exchanges could be embedded in the message semantic content at
egocentric level and then addressed to subjective agents for subjective externalisation. By
having different externalisations from different agent clusters I noted the importance of
proposing an intelligent system specific cluster model and architecture enabling a
subjective type of communication. These models and architectures can be used for the
design and development of intelligent agent clusters able to model subjective interface
tools to be utilised in knowledge exchange activities by intelligent systems users.
The analysis confirmed the exploration value of modelling and creating software agents
clusters with subjective communication capabilities and with internalised egocentric
communication properties with the ability to internalise representations of ontology
knowledge and subsequently to externalise internally stored knowledge representations to
other agents.
The analysis demonstrated that subjective qualities in artificial software agents’
communication exchanges could be embedded in the message semantic content at
egocentric level and then addressed to subjective agents for subjective externalisation.
Chapter 8: Conclusions
8.1 Research Background
This research investigation started with the realisation that software agents are a strongly
communicative section of the Intelligent Systems discipline. Software agents are currently
embedded in applications where the user interface is depicted as “intelligent” while the
Human to Agent Interface (HAI) that should foster communication between user and
software agent is not yet optimised to the level of two way subjective communications.
A number of considerations emerged in the preliminary investigation stages with the first
being that intelligence involves conceptual knowledge of a developmental nature and
shifting according to objective world changes so the research approach needed to be
“multidisciplinary” touching a range of disciplines such as Artificial Intelligence,
Information Sciences, Cognitive Sciences and Philosophy. The second important
consideration is that an intelligent entity develops and shifts knowledge through a verbal
and non-verbal range of communicative exchanges. A third consideration is that
cooperative power user applications currently embed software agent tools that lack the
subjective communicative aspects to distinguish what should be an artificial intelligent
entity from an automatic softbot. From these considerations emerged a realisation that
more investigative work in the area of software agent communication models could
improve the subjective communication model between humans and artificial intelligence
systems application tools. While intelligent software applications power users usually
“adapt” to the idiosyncrasies of their smart tools, nevertheless there are cognitive costs
related to the subjective nature of humans. Human learning and cognition is based on
internalising concepts of an objective environment and externalising the concepts in a
subjective way after a rational review in an egocentric communicative process. To have a
synchronous communication between a subjective human and an artificial intelligent
system, the artificial system should have some degree of subjectivity.
The project started by trying to investigate the characteristics of an artificial intelligence
model that could possess a basic degree of subjective communication properties to
perform intelligent communication patterns in a subjective fashion. Investigating
Artificial Intelligence (AI) tools required the involvement of a number of correlated
disciplines such as philosophy, cognition and information sciences. The challenging
problem of how intelligent system tools could be modelled to enable subjective
communication with conceptual transfer started to emerge. The design of software agent
architectures tended to follow a traditional tool design model envisaging a software agent
just as a software tool while we could consider a communicative software agent as a “tool
user” entity because it uses Agent Communication Language (ACL) as a tool. Entities
able to use tools to achieve goals are described as possessing various degree of
intelligence and if a software agent model is able to use communication tools in a
subjective way it may be considered functionally intelligent. Vygotsky’s Activity Theory
provided the theoretical framework for the investigation as it centred on areas of
developmental cognitive thinking and communication using language and speech as a
tool. Instead of analysing conceptual thinking separated from the action of
communicating it, Vygotsky’s theory of cognitive development demonstrated that
externalised communication involved conceptual material previously organised by inner
or egocentric speech prior to the externalisation process. Egocentric speech verbal
process facilitates the selection of essential conceptual thought from the non-essential.
Egocentric speech is also considered by Vygotsky to be an important factor in the
transition from internal thought to external communication. (Vygotsky 1986)
In Human-Computer Interaction (HCI), Vygotsky’s cognitive theories took the form of
Activity Theory (AT) where user actions are modelled as tool mediated activities where
users use a computer application as a tool. In my research investigation, I have used AT
as a framework, firstly because the theory was ideally suited to software agent
communication using “ACL language tools” to exchange messages with internalisation
and externalisation subjective properties and secondly because AT could also cater for
Human-Agent Interaction issues. Being the intelligent system model under research, a
“user of communication tool”, it was also suitable for modelling subjective traits
communication exchanges observed “internally” in the agent itself.
The multidisciplinary approach also involved the philosophy discipline and in particular
phenomenology as central to subjectivity and fitting with Vygotsky’s subjectivity,
objectivity and egocentric communication. The research theoretical study described
above was instrumental and challenging to investigate an intelligent system model able to
address Vygotsky’s AT with its aspects of internalised subject-object communication
activities mediated by language related tools. However, as concepts are communicated in
intelligent systems by ontology concepts embedded in a system communication language,
the ability for the model to switch ontology from the external environment to the internal
was necessary. Based on the model just described, I noted that two main questions were
1. What was the role that subjectivity could play in artificial intelligent systems
communication and learning?
2. Could the intelligent system model internal environment architecture achieve
subjective internalisation and egocentrism?
These research questions influenced the choice of a software agent system platform and
architecture suited for research tasks at hand. A range of factors weighed on the platform
choice: first that the software agent should be communicative, should use a standard
agent communication language and ontology as communication tool and secondly was
the ability to monitor and track the communication exchange both external and internal.
A more important determining factor was the ability to model Vygotsky’s theories of
egocentric speech that translated to a software agent environment signifying that an agent
should be able to send messages to “itself”.
A solution to address the factors above was found in a “virtual meta-agent cluster”
named ALMA (Agent Language Mediated Activity). All communicative activities inside
and outside the ALMA cluster could be monitored by “sniffing” and recorded. The
ALMA cluster platform embedded three JADE (Java Agent Development Environment)
software agents modelling Vygotsky’s Activity Theory with the functions of Objective
agent, Subjective agent and egocentric ALMA agent. The ALMA agent was intended to
be the cluster’s subjective core and the cluster taking its identity name. The ALMA three
agents’ virtual cluster was assembled to model a subjective intelligent system able to
internalise information and concepts relevant to external domain ontology and to
perform subjective egocentric communication using its own internal ontology.
During the research investigation, the concept of ALMA virtual agent cluster having an
external and internal ontology provided the opportunity to choose an ontology domain
suited for subjective applications relevant to legal environments. A basic legal ontology
LKIF (Legal Knowledge Interface Framework) was sourced from The University of
Amsterdam Leibniz centre AI and Law. However, the ALMA virtual cluster was made of
JADE agents that could not understand OWL ontologies. The ontology was translated in
JADE javabeans ontology format and the legal case used in the faculty of Law teaching
PiersonvPost was setup and merged to the LKIF ontology to obtain external domain
ontology able to be used in the model under test. With the same method I have created
an internal ontology named “Cogito” invisible from the external communication
exchanges but able to be accessed exclusively by the internal ALMA agent.
8.2 Research Conclusions
A single meta-agent cluster ALMA was used for the first round of tests aimed at the
model under investigation. The cluster contained an objective, a subjective and an
egocentric communication Alma agent. The design tested communication exchanges
between an external agent and the meta-agent cluster ALMA using the external domain
environment ontology specifically converted from OWL language to JADE java beans
format for the tests. Tests A1 and A2 were designed to assess the meta-agent cluster
architecture model by exploring the feasibility of objective internalisation achieved by
using a query ACL message from an external JADE agent to the Object agent in the
ALMA cluster. The Object agent internalised the message to ALMA internal agent that
then performed an egocentric communication message to itself and subsequently sent a
message to the Subject agent for externalisation to the external JADE agent. While the
tests A1 and A2 confirmed the proposed functional model, it was not sufficient to
demonstrate the expression of subjective qualities to a subjective agent. If unaware of the
fact that the model under test had a subjective model with internalisation and
externalisation properties the query-response mechanism could be seen as a common
standard software agent’s response behaviour. Human users are inherently subjective
entities and, as such, can pick easily if an intelligent system has subjective properties or if
it is just a standard software agent. The determination of an entity as subjective is
performed by considering the conceptual content of subjective knowledge externalisation
during the communication exchange. Only another entity with subjective properties can
achieve that. Subjective quality in an intelligent system can be achieved by egocentric
communication patterns that can transpire during the externalisation processes in the
conceptual content of messages. Any successful experiment on artificial intelligent
systems’ subjectivity required firstly, an egocentric communication inside the cluster
model and secondly to provide a subjective externalisation with concepts that could
digress from the external domain knowledge showing subjective conceptual qualities.
The single meta-agent cluster architectural model used in tests A1 and A2 was intended
to test a basic ALMA subjective intelligent system model functionality while tests B1 and
B2 used two subjective clusters ALMA and ALMA2 to test subjective exchanges between
meta-agent clusters. A1 test demonstrated that Vygotsky’s internalisation and
externalisation communicative flow could be used in an artificial intelligent system model.
More significant were results demonstrating the achievement of egocentric
communication internally in the model sending and receiving messages addressed to the
agent self thus validating the possibility of building software meta-agents architectures
able to conceptualise learning and subjective externalisation. Both A1 and A2 tests
validated the functionality of the proposed software agent cluster architectural model
providing the feasibility of a Vygotsky’s internalisation and an egocentric communication
in artificial intelligent systems. The tests and the model provided the interesting
conclusion that subjectivity is enabled by the egocentric communication achieved prior to
the process of externalisation first and second by the externalised message conceptual
content. To investigate the role of subjective communication in intelligent systems
modelling of communication exchanges between two artificial subjective agent clusters
then become necessary.
I therefore set up an agent architectural model with two clusters performing
communicative tests B1 and B2 where, in test B1, I have tested a knowledge “escalation”
model where a subjective agent cluster ALMA receive an external message and tried to
escalate the request to another agent cluster ALMA2 to possibly obtain further
conceptual information. In this test ALMA2 is a subjective cluster providing a subjective
externalisation message to both ALMA and an external software agent DA1. Note that
DA1 is not a subjective agent cluster but a software agent. This test demonstrated that
subjective externalisation became part of the objective world when sent out in the
environment and that only a receiving subjective entity as in this case an agent cluster, will
recognise the subjective nature of the message from its content. In conclusion, it was test
B2 that provided an improved understanding of the possible role of subjectivity in
artificial intelligent systems as expressed in the first research query.
The test involved the subjective evaluation model performed by two subjective agent
clusters ALMA and ALMA2. From the resulting communicative exchange we can
observe that a subjective evaluation communication exchange is performed with a cluster
returning a positive confirmation message and the other cluster a non confirmation
message related to the same query. The B1 and B2 tests confirmed that specifically
designed artificial intelligent system architectures such as ALMA can provide modelling
of different subjective situations such as knowledge escalation and knowledge
differentiation. This causal relation between egocentric communication exchanges and
subjective externalisation properties obtained from the research investigations is
important to answer the research questions on subjectivity and the possibility of
egocentric communication in artificial agent systems.
The constructed artificial subjective model demonstrated that subjectivity can have an
“ontological shift” role in the conceptual learning process and a “subjective switch” role
in determining the adequate externalisation of internalised concepts. The “ontological
shift” role is achieved by the processes of internalisation, egocentric communication and
the use of an internal ontology that can store and change conceptual objects and their
properties. The “subjective switch” role is achieved by using the egocentric
communication and the requirement to externalise the content of acquired internal
ontology concepts to an external objective domain.
We can conclude that subjectivity has the role of enabling ontological shifts conducive to
conceptual learning in communicating artificial intelligent systems provided the
communicative exchange is performed between artificial intelligent systems possessing
basic egocentric communication capabilities together with subjective properties involving
concept’s externalisation and internalisation. Having established on the modelling tests
that subjective qualities are derived by egocentric communication capabilities, we
conclude that egocentric exchanges are important in determining the externalised
subjective properties that intelligent systems can use in subjective communication.
Artificial intelligent systems egocentric communication is an area that deserves to be
followed up in future research efforts.
8.3 Research Contribution
This research contributes to improve the design and modelling process of intelligent
agent systems able to internalise and subjectively externalise domain knowledge
improving human-intelligent systems interaction and communication. This research
highlights the possibility of using Vygotsky’s theories of cognition and learning in the
field of Human-Agent interaction. The research experiments and the subsequent analysis
found that artificial internal egocentric communication is feasible and this can be
determinant for the modelling and the creation of artificial subjective systems
contributing to the improvement of artificial intelligent systems communication with
users in the area of Human-Agent interaction.
This result has been achieved by modelling with the experiment the symmetric and
asymmetric communication exchanges between intelligent systems and by defining a new
experimental intelligent system model addressing the problem related to recording and
analysing the internal communication process inside an artificial subjective entity. This is
an area that cognitive sciences will find quite useful providing the possibility of using
artificial models in testing theories of egocentric communication and internalisation
aspects of learning.
The current research will contribute also to solve the problem of robustly modelling
cooperative software agent communicative acts using Activity Theory (AT) when applied
to business intelligence and knowledge mining cooperation’s in corporate and
Government enterprise environments.
Rather than focusing only on objective agents external communicative activities, this
research investigated a model able to investigate subjective internal communicative
activities with the aim of determining the role that subjectivity has in the areas of
cooperation, internal agent architecture, agent internal knowledge and self identification
in the environment.
Philosophy of mind and artificial intelligence disciplines may also find the model useful in
investigations related to the area of artificial consciousness. In the area of artificial
intelligent systems communication, this research has tried to clarify issues of subjectivity
and egocentric capabilities in communicative exchanges among intelligent systems.
While system engineering research is active in physical network agent cooperation and
communication in the Internet environment, a multidisciplinary study of agent
internalization of knowledge using subjective models has successfully been attempted.
Enhancing the domain knowledge in the area of agent identification and re-identification
of objects external to the agent self and therefore improve the introspection process and
agent concept of the self.
The research will be also extremely important in the area of Internet knowledge mining
and e-Commerce where many subjective agents can communicate and collaborate with
human users for the achievement of common goals. Being able to identify private goals
against other agents behaviour will then facilitate and expedite knowledge mining and
processing. The modelling of agent internal processes took a defined approach with
internalisation of conceptual knowledge and ontology as main areas of research.
Subjective agent internalization of identified search routes together with re-identification
of familiar ones can expedite, to a great extent, massive internet searches by eliminating
duplication of responses in multiple agent searches; it could also eliminate cognitive
overload in computer users when they examine searches results.
Intelligent Systems expressing subjective traits when communicating with humans will be
able to use communicative tools cater for introspective activity capable of learning and
internalise external environment data and subjectively externalise internal knowledge. The
design of the subjective intelligent agent architecture ALMA provides software agent
subjective properties to be applied in the areas of internet e-business and Business
Intelligence. In e-Commerce environments a collaborative agent with introspection will
be invaluable in particular in the areas of e-Procurements and e-Auctions where the agent
can identify and compare his internal biddings with external agents bidding activities.
Identification and re-identification of e-Commerce transactions are also important for the
analysis and monitoring of good transaction processes with the related control of non
conforming corporate transactions and Internet frauds.
In the area of Law enforcement and Defence intelligence, the subjective identification
and re-identification of intelligence reports, signals or message patterns and the relevant
knowledge internalization makes the proposed agent under research an ideal solution to
this area.
This research will contribute also to the area of Business Intelligence where the capability
for multiple subjective intelligent agents to capture, collate and internalize data and
subsequently organise subjective knowledge using inductive inferences is necessary in
order to build knowledge bases related to external events. Artificial subjective models can
also be used to investigate philosophical theories in the areas of phenomenology, identity
and consciousness.
8.4 Future Research
This research focused on an agents cluster model focused on defining a model able to
increase understanding of the possible role of subjectivity and egocentric communication
in artificial intelligent systems. The research generated an incentive to continue my
multidisciplinary investigative work in an area related to subjectivity and subjective
knowledge management systems. During the research period further subjective intelligent
systems areas requiring a deeper enquiry have been identified, however, to maintain focus
on the thesis research questions I had to set these investigations aside for future research
activities. Two promising such research interests arose from the ALMA subjective model
testing and originated from the relationship between the model egocentric
communication and the subjective externalisation process. By establishing in the current
research investigation that an artificial Intelligent System model can perform egocentric
communication, further research investigation is required to design a “subjective” KBS
necessary to adequately control the “switching” transition from egocentric
communication to subjective externalisation. The second promising topic for further
investigative research is the area of “Qui custodiat custodes”, that is a Latin expression
for “Who controls the controller”? While a subjective KBS could control the switching
transition from egocentric communication mode to subjective externalisation mode, a
further “switching” function is necessary from egocentric communication to internalised
learning and recalling mode. This extended egocentric function should be researched and
modelled as an “intelligent sniffer” able to monitor all internalised messages and liaise
with the subjective KBS switch. Further research in enhanced ALMA models will assist
new investigations in the areas of Intelligent Systems cognition and learning, Cognitive
Sciences and in particular in the Philosophy of Mind area of Consciousness. The research
field of artificial intelligence could make use of artificial subjective models similar to
ALMA meta-agent clusters to increase knowledge in advanced research areas such as
artificial consciousness and Human- Subjective Intelligent Systems interaction.
Adams, W 1994 Machine Consciousness, Plausible Idea or Semantic Distortion?
Journal of Consciousness Studies, 11 (9): 45-56.
Andre, E, Muller, J & Rist, T 1997, Web Persona: A Lifelike Presentation Agent for
the World Wide Web, Proceedings of IJCAI-97, Workshop on Animated Interface
Agents, Nagoya, Japan.
Artikis, A, Sergot, M and PittJ, 2009 Specifying norm-governed computational
societies, ACM Transactions on Computational Logic, 10(1), 2009.
Arvola, M 2010, Interaction design qualities: theory and practice,
Proceedings of the 6th Nordic Conference on Human-Computer Interaction:
Extending Boundaries, October 16–20, 2010, Reykjavik, Iceland
Austin, JL, 1965. How to do Things with Words. Oxford University Press, NY.
Baber, C, 2003, Cognition and Tool Use: Forms of Engagement in Human and
Animal Use of Tools, Taylor & Francis.
Bass, L, Clements, P & Kazman, R 2003, Software Architectures in Practice
(2ndEdition), Reading (MA), Addison –Wesley.
Bergamaschi, S, 1997, Estraction of Information from highly Heterogeneous
Sources of Textual Data, in P Kandzia and M Klush (Eds.) Cooperative Information
Agents, Proc. Workshop CIA-97, Kiel Germany, Springer LNAI, Vol 1202, 1997.
Bergenti, F, Rimassa, G, Poggi, A & Turci, P 2002, Middleware and Programming
Support for Agent Systems, in Proceedings of the 2nd International Symposium from
Agent Theory to Agent Implementation, Vienna, pp. 617-622.
Beydoun, G, Low, G, Henderson-Sellers, B, Mouratidis, H, Gomez-Sanz, J, Pavon, J
,Cesar Gonzalez-Perez, C 2009, FAML: A Generic Metamodel for MAS
Development, IEEE Transactions on Software Engineering, v.35 n.6, p.841-863,
November 2009.
Blass, A 1990, Infinitary Combinatorics and Modal Logic, Journal of Symbolic
Logic,vol. 55 n.2 March 1990 ppg. 761-778.
Boer, A, Hoekstra R, & Winkels R 2002, MetaLex: Legislation in XML. In
Proceedings of JURIX 2002: Legal Knowledge and Information System, 1–10.
Boer, A, Winkels R, & Vitali, F 2007, XML Standards for Law: MetaLex and LKIF,
in Proceedings of JURIX 2007, Amsterdam, IOS.
Bonabeu, E, Dorigo, M & Theraulaz, G 1999, Swarm Intelligence.From Natural to
Artificial System, Oxford University Press, UK.
Breuker, J, Boer, A, Hoekstra, R, & van den Berg, K 2006, Developing content for
LKIF: Ontologies and frameworks for legal reasoning, in TM van Engers (ed.) Legal
Knowledge and Information Systems. JURIX 2006: Nineteenth Annual Conference,
volume 152 of Frontiers in Artificial Intelligence and Applications.
Breuker, J & Hoekstra, R 2004a, Core concepts of law: taking common-sense
seriously, in A Varzi & L Vieu (eds.), Proceedings of Formal Ontologies in
Information Systems (FOIS-2004), pages 210–221.IOS-Press.
Breuker, J, Valente, A & Winkels, R 2004b, Legal Ontologies in Knowledge Engineering
and Information Management, Artificial Intelligence and Law, 12:241–277.
Busetta, P, Hodgson, A & Ronnquist, R 1999, Specification of Coordinated Agent
Behaviour, IJCAI ’99 Workshop on Team Behavior & Plan Recognition, pp 75-81.
Cabri, G, Leonardi, L & Zambonelli, F 2002, Engineering Mobile Agent Applications via
context-dependent Coordination, IEEE Transaction on Software Engineering, vol. 28
(11), 1034-1051.
Castelfranchi, C, & Werner E 1994, Artificial Social Systems, LNAI, Springer.
Chalmers, DJ 1996, The Conscious Mind, NY Oxford University Press.
Chalmers, DJ 2000, What is a Neural Correlate of Consciousness? T Metzinger ed.
Neural Correlates of Consciousness, MIT Press.
Chalmers, DJ 2007 Contemporary Philosophy of Mind, annotated bibliography. ANU
Chavez, A, Maes, P 1996, Kasbah: An agent marketplace for buying and selling goods.
Proceedings of the First International Conference on the Practical Applications of
Intelligent Agents and Multi Agents Technology (PAAM-96), Pages 75-90, April 1996.
Chomsky, N 1975, Reflections of Language, New York, Random House.
Chomsky, N 1980, Rules and Representations, New York, Columbia University Press.
Ciancarini, P and Wooldridge, M 2001, Agent-Oriented Software Engineering in
Proceedings of the 1st International Workshop on Agent-Oriented Software Engineering.
Vol. 1957 of LNCS, Springer Verlag, pp. 1-24.
Clark A & Chalmers, D 1998.The Extended Mind, Analysis, 58: 10-23.
Cohen, PR & Levesque, HJ 1990, Persistence, Intention and Commitment, in Intentions
in Communication, Cohen et al. (eds.), MIT Press, Cambridge, Mass. Pp. 33-69.
Cohen, PR, Levesque, HJ 1995, Communicative Actions for Artificial Agents,
Proceedings of the First International Conference on Mlulti-Agent Systems, AAAI Press,
San Francisco, June 1995.
Coliva, A, 2009 Was Wittgenstein an Epistemic Relativist? In Philosophical
Investigations, vol. 33, ppg 1-23, January 2010
Cordeschi, R. 2010, Which Kind of Machine Consciousness? International Journal of
Machine Consciousness, vol 2, issue 1-2010, pp 31-33.
Damasio, AR 2000, A Neurobiology of Consciousness, in T Metzinger eds. Neural
Correlates of Consciousness, MIT Press.
Dauntehahn, K & Numaoka, C 1998, International Journal on Applied Artificial
Intelligence, Special issue on Socially Intelligent Agents, vol. 12 (7-8), 1.
Dennett, DC 1991, Consciousness Explained, Boston, Little Brown.
Dennett, DC, 1994, Consciousness in Human and Robot Minds, Proceedings from
IIAS Symposium on Cognition, Computation and Consciousness. Kyoto, Japan,
September 1994.
Doran, JE, Franklin, S, Jennings, NR & Norman, TJ 1997, On Cooperation in MultiAgents Systems, The Knowledge Engineering Review, Vol. 12, (3).
Edelman, G & Tononi, G 2000, A Universe of Consciousness, Basic Books. USA
Edmond, D & Papazoglou, MP 1998, Reflection is the essence of cooperation. In
MP Papazoglou and G Schlaegeter (Eds.), Cooperative Information systems: Trends
and Directions, Academic Press, pp.233-262.
Engster, P, Feiber, PA, Guerraoni, R & Kermarrec, A 2003. The Many Faces of
Publish/Subscribe, ACM Computing Surveys vol. 35 (2), pp. 114-131.
Etzioni, O & Weld, D 1996, A SoftBot based interface to the Internet, ACM
Communications, vol. 37(7), 1996.
Euzenat, J & Shvaiko, P 2007, Ontology Matching, Springer-Verlag, Berlin.
Evans, G 1982, Varieties of Reference. Ed. J. McDowell. Oxford, University Press.
Finin, T, McKay, D, Fritzson, R & McEntire, R 1994, KCaML, An Information
and Knowledge Exchange Protocol, K Fuchi & T Yokoi (Eds.), Knowledge
Building and Knowledge Sharing, Ohmsha and IOS Press.
See also:
FIPA, 1997, Agent Communication Language Specifications
FIPA, 2002, Agent Communication Language Specifications
Forrest, P 2008, "The Identity of Indiscernibles", The Stanford Encyclopedia of
Philosophy (Fall 2008 Edition), Edward N. Zalta (ed.)
Franklin, S & Gaesser, A 1996, It is an Agent or just a program? A taxonomy for
autonomous agents, Proc. 3rd International Workshop on Agent Theories,
Architectures and Languages (ATAL-96), Springer, LNAI.
Garcia-Molina, H, et al. 1995, The TSIMMIS Approach to Mediation: Data Models
and Languages, Proceedings Workshop NGITS.
Gehmeyr, A, Muller, J, & Schappert, A 1998, Mobile Information agents on the
Web, Klush & Weiss Eds., Cooperative Information Agents II, Proc. International
Workshop CIA-98, Springer, LNAI, Vol. 1435, 1998.
Gensereth, MG, & Katchpel, SP 1994, Software Agents, Communications of ACM
Vol 37, N.7, 1994 pp. 48-53, 147.
Gervais, M, Gomez, J and Weiss, G 2004, A Survey on Agent-Oriented Software
Engineering Researches, in Methodologies and Software Engineering for Agent
Systems, Kluver, NY.
Gibson, KR & Ingold, T 1993, Tools, Language and Cognition in Human Evolution,
Cambridge University Press
Gobbin, R 1998a, The role of cultural fitness in user resistance to information
technology tools, Masters Thesis, University of Canberra.
Gobbin, R 1998b, Adoption or Rejection: Information systems and their cultural
fitness, Information Systems and Activity Theory: Tools in Context. H. Hasan, E.
Gould & P. Hyland (Eds.) Wollongong University Press.
Gobbin, R 1998c, The role of cultural fitness in user resistance to information
technology tools, Interacting with Computers, n.9, 275-285. Elsevier Science UK.
Gobbin, R, Jentzsch, R & Mohammadian, M 2004, Using agent subjective
properties in modelling multiple agent communication activities, Conference on
Human Computer Interaction 2004 Wollongong.
Gobbin, R 2004, Intentional Subjective Properties: Communication Modeling Tools
for Online Dispute Resolution Software Agents, IAWTIC 2004 Conference
Proceedings, Australia
Gobbin R 2006, The Role of Semantic Tools in the Construction of Subjective
Meaning in Multi-agents Communication Systems, IEEE IAWTIC 2005 Conference
Proceedings, Austria.
Gobbin, R, 2004 The Application of Intentional Subjective Properties and Mediated
Communication Tools to Software Agents in Online Dispute Resolution Environments.
Australasian Journal of Information Sciences Vol. 12 n.1 September 2004
Goldspink, C, 2009, Agent Cognitive Capability and Orders of Emergence, in:
Proceedings of AISB Convention, Communication, Interaction and Social
Intelligence Aberdeen, UK.
Gomez-Sanz, J & Pavon, J 2003, Agent Oriented Software engineering with
INGENIAS, In: Proceedings of the 3rd Central and Eastern Europe Conference on
Multiagents Systems, LNCS, Vol. 2691, pp. 394-403, Springer-Verlag.
Grice, P 1989, Studies in the Way of Words, Harvard University Press, Cambridge,
Gruber, TR, 1993, A translation Approach to Portable Ontology Specifications,
Knowledge Acquisition, Vol. 2, 1993, pp.199-220.
Grudin, J 1990, The Computer Reaches Out: The Historical Continuity of User
Interface Design, Proceeding of CHI '90, ACMSIGCHI Conference, SEATTLE.
Grudin, J & Gentner, DR 1996, Design models For Computer Human Interfaces,
COMPUTER, June 1996, 29, 6, 28-35, IEEE Press.
Guarino, N, Carrara, M & Giaretta, P 1994, An Ontology of Meta-level Categories,
Journal of Knowledge Representation and Reasoning: Proceedings of the Fourth
International conference (KR94), Morgan Kaufmann, San Mateo, CA.
Guarino, N & Welty, C 2000, Ontological Analysis of Taxonomic Relationships, A
Laender & Storey, V eds., Proocedings of ER-2000: The International Conference
on Conceptual Modeling, October 2000, Springer-Verlag, LNCS.
Guerin, F & Pitt, J 2001, Denotational Semantics for Agents Communication
Language, Proceedings from the Fifth International Conference on Autonomous
Agents, Montreal Quebec, Canada, pp. 497-504.
Hachicha, H, Loukil, A & Ghedira, K, 2009, MA-UML: a conceptual approach for
mobile agents' modelling, International Journal of Agent-Oriented Software
Engineering, v.3 n.2/3, p.277-305, March 2009
Hanhs, MN, Stephens LM 2002, Multiagent Systems and Societies of Agents, in G
Weiss (ed.) Multiagent Systems: A Modern Approach to Distributed Artificial
Intelligence, MIT Press USA.
Hendler, J 2007, Where Are All the Intelligent Agents? IEEE Intelligent Systems,
n.7, 2-3.
Hewitt, C & Inman, J 1991, DAI Betwixt and Between: From "Intelligent Agents" to
Open Systems Science, IEEE Transactions on Systems, Man, and Cybernetics, Nov.
/Dec. 1991.
Hubner, J F, Vercouter, L & O. Boissier. 2008. Instrumenting multi-agent
organisations with artifacts to support reputation processes. In COIN 2008, pages
96–110, Berlin, Heidelberg, 2009, Springer-Verlag
Hutchins, E 2010, Cognitive Ecology in Topics in Cognitive Science, vol 2 n.4,
pages 705–715, October 2010.
Ilarri, S., Mena, E.& Illarramendi, A.2008, Using cooperative mobile agents to
monitor distributed and dynamic environments. Information Sciences, vol 178:
Jacquette, D 2002, Ontology, Acumen Publishing, UK
Jennings, NR 2001, An Agent-based Approach for Building Complex Software
Systems, Communication of ACM ,44(4), 35-41.
Jentzch, R & Gobbin, R 2003, A Cooperative Communicative Intelligent Agent
Model for E-Commerce, Managing E-commerce and Mobile Computing
Technologies, J Mariga, (ed.), IRM Press US
Jentzsch, R, Masoud M & Gobbin, R, 2004, A Framework for using RFID in a
Hospital, IAWTC 2004 Conference Proceedings.
Juan, T, Pierce, A, & Sterling, L 2002, ROADMAP: Extending the Gaia
Methodology for Complex Open Systems, Proceedings of the 1st AC Joint
Conference on Autonomous Agents and Multi-Agents Systems, Bologna, ACM Press
pp 3-10.
Kaptelinin, V 1992, Human Computer Interaction in Context: The Activity Theory
perspective, in Proceedings, East-West Human Computer Interaction Conference, St
Petersburg, 1992.
Kaptelinin, V 1996, Computer Mediated Activity, Context and Consciousness, B
Nardi (ed.), MIT Press, Cambridge.
Keil, F 1979, Semantic and Conceptual Development: An Ontological Perspective,
Cambridge, MA: Harvard University Press.
Kennedy, M. 2010, Naive realism and experiential evidence, Proceedings of the
Aristotelian Society 110 (1):77-109.
Kempson, RM 1977, Semantic Theory. Cambridge University Press, UK.
Kinny, D & Georgeff, M 1996, A Methodology and Modeling Technique for Systems of
BDI Agents, in Proceedings of the 1st Workshop on Modeling Autonomous Agents in a
Multi-agent World, Vol. 1038 of LNAI pp. 56-71,Springer-Verlag.
Klush, M 2000, Intelligent Information Agents: Agent Based Information Discovery
and Management on the Internet, Springer Verlag.
Knoblock, CA, Arens, Y & Hsu, CN 1994, Cooperating Agents for Information
Retrieval, Proceedings of 2nd International Conference on Cooperative Information
Systems, ppg. 122-133, Toronto, Canada
Kolp, M, Giorgini, P & Mylopoulos, J 2002, A Goal Based Organisational
Perspectives on Multi-agent Architectures, in: Intelligent Agents VIII: Agents
Theories Architectures and Languages, Vol. 2333, LNAI, Springer-Verlag, ppg.
Kuutti, K 2009, Re-mediation-A Potentially Useful New Concept for Design
Research, in Proc. IASDR 2009, Korean Society of Design Science (2009).
Kuutti, K 1991, Activity Theory and its Applications to Information Systems, in HE
Nissen (ed.), Information Systems Research, Amsterdam, Elsevier Science.
Kuutti, K 1996, Activity Theory as a Potential Framework for HCI Research,
Context and Consciousness, B Nardi, (ed.) MIT Press, Cambridge.
Labrou, Y & Finin, T 1998, Semantics for an Agent Communication Language,
Agent Theories, Architectures and Languages IV, M Woolridge, et al. (eds.),
Lecture Notes in Artificial Intelligence, Springer-Verlag, Berlin.
Labrou, Y, Finin, T & Peng, Y 1999, Agent Communication Languages: The
Current Landscape, IEEE Intelligent Systems March/April 1999, ppg.45-52.
Laera, L, Tamma, V, Euzenat, J, Bench-Capon, T & Payne, T 2006, Reaching
Agreement over Ontology Alignment, In Proc. 5th International semantic web
Conference (ISWC), Lecture Notes in Computer Science, vol 4273, ppg.371-384,
Athens Ga US.
Le Dantec, CA & Do, EY., 2009. The mechanisms of value transfer in design
meetings. Design Studies, vol.30 n. 2, 119–137
Le Dantec, CA 2010, Situating Design as social Creation and Cultural Cognition, in
Co-Design, vol. 6 n.4, December 2010, 207–224.Lock, AJ & Peters, CR 1996,
Social Relations, Communication and Cognition, Handbook of Human Symbolic
Evolution, Clarendon Press Oxford.
Luger, GF & Stubblefield, WA 1998, Artificial Intelligence: Structures and
Strategies for complex Problem Solving. Addison Wesley.
Macal, CM & North, MJ, 2010, Tutorial on Agent-based Modelling and Simulation,
Journal of Simulation vol 4, 151–162
Mandik, P 2001, Mental Representation and the Subjectivity of Consciousness,
Philosophical Psychology, 14 (2):179-202.
Mandik, P 2002, Synthetic Neuroethology, Metaphilosophy, 33 (1-2): 11-29,
reprinted in CyberPhilosophy: The Intersection of Philosophy and Computing, JH
Moor & TW Bynum, (eds.), Oxford: Blackwell, 2002.
Mandik, P 2008, The Neural Accomplishment of Objectivity, in P Poirier & L
Faucher (eds.), Des Neurones a La Philosophie: Neurophilosophie Et Philosophie
Des Neurosciences, Éditions Syllepse.
Mandik, P 2009, The Neurophilosophy of Subjectivity, in J Bickle (ed.), Oxford
Handbook of Philosophy and Neuroscience, Oxford University Press.
Margolis, E & Laurence, S 2007, The Ontology of Concepts: Abstract Objects or
Mental Representations, in NOUS, 41:4, ppg.561-593, Blackwell Publishing Inc.
McBurney, P & Parsons, S 2009, Dialogue games for agent argumentation, Rahwan,
I & Simari, G (eds), Argumentation in Artificial Intelligence,chapter 13, pp. 261280, Springer, Berlin
Metzinger, T 2000, Neural Correlates of Consciousness, MIT Press, Cambridge
Metzinger, T 2003, Being No One, MIT Press, Cambridge
Mithin, S 1996, The Prehistory of the Mind: The Cognitive origins of Art, Religion
and Science, London, Thames & Hudson.
Moses, Y & Tennenholtz, M 1995, Artificial Social Systems, Computers and
Artificial Intelligence, 14(3), 533-562.
Mouratidis, H & Huget, M-P 2008, Special issue: Modelling Languages for Agent
Systems, International Journal Agent-Oriented Software Eng., 2(4), 379-474
Nagao, K, & Takeuchi, A 1994, Social Interaction: Multimodal conversation with
social agents, Proceedings of AAAI-94 Conference, Seattle.
Nagel, T 1974, What Is It Like to be a Bat, Philosophical Review 83 (1974), pp.435450
Nardi, BA 1996, Studying Context, Context and Consciousness, MIT Press,
Nass, CS, Tauber, J & Ellen, R 1994, Computers are social actors, Proceedings of
CHI-94 Conference, Boston.
Neches, R 1991, Enabling Technology for Knowledge Sharing, AI Magazine,
Vol.12, No.3, Fall, 1991, pp. 36-56.
Nodine, M, Bohrer, W and Ngu, AHN 1999, Semantic Brokering over Dynamic
Heterogeneous Data Sources in InfoSleuth, Proceedings of International Conference
on Data Engineering, ICDE-99.
Noriega, P 1998, Agent-Mediated auctions: The Fishmarket Metaphor, PhD
Doctoral Thesis from the Universitat Autonoma de Barcelona.
Odell, J 2000, Agent Technology, OMG Document 00-09-01, OMG Agents interest
Group, September 2000.
Oderberg, DS 2009 The non-identity of the categorical and the dispositional
Analysis vol 69:677-684.
Oliva, E, Viroli, M,Omicini, A & McBurney, P. 2009. Argumentation and artifact
for dialogue support, in I. Rahwan and P. Moraitis (Eds). Argumentation in MultiAgent Systems, vol. 5384 of Lecture Notes in Computer Science, Springer
Oliva, E, McBurney, P, Omicini A & Viroli, M 2010, Argumentation and artefacts
for Negotiation Support in International Journal of Artificial Intelligence, Spring
2010, Vol 4, n S10, CESER Publications
Omicini, A 2001, SODA: Societies and Infrastructures in the Analysis and Design of
agent-based systems, In Proceedings of the 1st International Workshop on Agent
Oriented Software Engineering, Vol. 1957, LNCS, Springer Verlag, pp. 185-194.
Omicini, A & Rimassa, G 2004, Integrating Objective and Subjective Coordination
in Multi-Agents Systems, Proceedings of the 2004 ACM Symposium on Applied
Computing, Nicosia, Cyprus 2004, pp.449-455, ACM Press.
Omicini, A & Ossoswski, S 2003, Objective versus Subjective Coordination in the
Engineering of Agent Systems, LNCS Springer Berlin Vol. 2586 pp. 179-202.
Omicini, A, Ricci, A & Viroli, M 2006, Agens Faber: Toward a Theory of Artifacts
for MAS. Electronic Notes in Theoretical Computer Sciences, 150(3): 21-36.
Omicini, A , Ricci, A & Viroli, M 2008 Artefacts in the A&A meta-model for multi
agent systems in Autonomous Agent Multi-Agent systems(2008) 17:432-456
Pariente, JC 1985, "Grammaire et Logique." In L'analyse du langage à Port-Royal:
six études logico-grammaticales. Paris: Les Editions de Minuit.
Parker, ST 1993, Higher intelligence, propositional language and culture as
adaptation for planning, in Tools, Language and Cognition in Human Evolution, K
Gibson & T Ingold (eds.), Cambridge University Press UK.
Parunak, H 1997, Go to the Ant: Engineering principles from Natural Agent
Systems, Annals of Operational Research 75, pp.69-101.
Patil, RS 1997, The DARPA Knowledge sharing Effort: Progress Report, Readings
in Agents, M Huhns and M Singh (eds.), Morgan Kaufmann.
Piaget, J 1954, The Construction of Reality in the Child, New York Basic Books.
Peirce, C S 1898, Reasoning and the Logic of Things, The Cambridge Conferences,
Lectures of 1898, K.L. Ketner ed. Harward University Press, Cambridge MA 1992.
Perlovsky, LI, 2006, Modelling Field Theory of Higher Cognitive Functions, Chapter in
Artificial Cognition Systems, Eds. A. Loula, R. Gudwin, J. Queiroz. Idea Group, Hershey,
PA, pp.64-105
Povinelli, DJ 2001, The Self: Elevated in consciousness and extended in time, in K
Skene & C Moore,(eds.) The Development of the extended self in pre-school
children: Theory and Research, Pp. 73-94, Cambridge University Press.
Preston, B 1998, Cognition and tool use, Mind and Language, 13 (4):513 -547.
Revonsuo, A 2000, Prospects for a Scientific Research Program on Consciousness,
in T Metzinger (ed.) Nervous Correlates of Consciousness, MIT Press.
Rahwan, I, Iatson, K & Tohme, F 2009, A characterisation of strategy-proofness for
grounded argumentation semantics, Proceedings of the 21st International Joint
Conference on Artificial Intelligence (IJCAI2009), Pasadena, CA, USA
Revonsuo, A 2006, Inner Presence: Consciousness as a Biological Phenomenon,
MIT Press, Cambridge.
Rey, G 1997, A question about Consciousness, in Block, Flanagan and Guzeldere
(eds.) The Nature of Consciousness: Philosophical Debates. Cambridge, MA: The
MIT Press pp. 461-482.
Ricci, A, Omicini, A & Denti, E 2003, Activity Theory as a Framework for MAS
Coordination, LNCS,Springer, Vol. 2577 pp. 295-394, Berlin.
Ricoeur, P 1992, Yourself as Another, University of Chicago Press USA.
Rissland, E, Ashley, K, & Loui, R 2003, AI and Law: A Fruitful Synergy, Artificial
Intelligence, Vol.150, pp.1-15.
Saariluoma, P & Nevala, K. 2007, From Concepts to Design Ontologies, in
"Proceedings of the The First International Workshop on Semantic Web and Web
2.0 in Architectural, Product and Engineering Design" CEUR Workshop
proceedings Vol 294 Busan, Korea 11-11-2007
Sartor, G et al. 2008, Legal Informatics and Management of Legislative Documents,
in Global Centre for ICT in Parliament, Working Paper n.2 32(2):121-129.
Shoham, Y & Tennenholtz M 1995, Social Laws for Artificial Agent Societies: OffLine Design, Artificial Intelligence, 73.
Sowa, JF 2001, Knowledge Representations, Brooks/Coles, USA
Sperber, D 1994, Understanding Verbal Understanding, in J Khalfa (ed.) What is
Intelligence?, Cambridge University Press, UK.
Sperber, D 2000, Meta-representations in an evolutionary perspective, in Dan
Sperber ed. Meta-representations: A Multidisciplinary perspective, Oxford
University Press, UK.
Sperber, D & Wilson, D 1995, Relevance: Communication and Cognition,
Blackwell Oxford.
St. Amant, R & Wood, A 2005, Tool Use for Autonomous Agents, AAAI
Conference 2005, USA.
Sterenly, K 2003, Thought in a Hostile world: The Evolution of Human Cognition,
Blackwell Oxford.
Strawson, G 2008 The identity of the categorical and the dispositional,
Analysis, vol 68:271–82.
Strawson, PF 1956, The Bounds of Sense: An Essay on Kant's Critique of Pure
Reason, Methuen London.
Strawson, Peter F. (1990) Individuals: An Essay in Descriptive Metaphysics,
Rutledge London.
Turing, AM 1950, Computer Machinery and Intelligence, Mind, 59, 433-460.
Van der Hoek, W & Wooldridge, M 2003, Towards a Logic of Rational Agency,
Logic Journal of IGPL, 11 (2), 135-160
Velan K 2009 Modelling Bidders in Sequential Automated Auctions,
The Computer Journal (2010) vol 53:208-218.
Vico G 1988, On the Most Ancient wisdom of the Italians: Unearthed from the
origins of Latin Language, (L.M. Palmer translation) Cornell University Press.
Vygotsky, LS 1978, Minds in Society, Harvard University Press, Cambridge.
Vygotsky, LS 1986, Thought and Language, MIT Press, Cambridge.
Zahavi, D, 2005, Subjectivity and Selfhood, MIT Press, Cambridge
Zambonelli, F, Jennings, N & Wooldridge, M 2003, Developing Multiagent
Systems: The Gaia Methodology, ACM Transactions on Software Engineering and
Methodology, 12 (3), 417-470.
Zambonelli, F & Omicini, A 2004, Challenges and Research Direction in AgentOriented Software Engineering, Kluvier Academic Publishers, Nederland.
Zhang, W, Serban, C and Minsky, N. Establishing global properties of multi-agent
systems via local laws.In D. Weyns, editor, Environments for Multiagent Systems
III, volume LNAI 4389, Springer, 2007.
Wang, Y 2010, Abstract intelligence and cognitive robots, Paladyn, Journal of
Behavioural Robotics, vol 1, n1, Springer
Wegner, DM 2002, The Illusion of Conscious Will, Cambridge, MA: MIT Press.
Weiss, G 2001, Multiagent Systems, MIT Press USA.
Wiederhold, G 1994 Interoperation, Mediation and Ontologies, Proceedings of
International Workshop on Heterogeneous cooperative Knowledge Bases, Tokyo
Wittgenstein, L 1958, The Blue and Brown Book, Harper & Row, NY.
Wood, M, DeLoach, A & Sparkman, C 2001, Multiagents Systems Engineering,
International Journal of Software Engineering and Knowledge Engineering, 11(3),
Wooldridge, M, Jennings, R & Kinny, D 2000, The Gaia Methodology for Agentoriented Analysis and Design, Journal of Autonomous Agents and Multiagent
Systems 3 (3), 285-312.
Wurman, PR, Wellman, MP & Walsh, WE 1998, The Michigan Internet
AuctionBot: A Configurable Auction Server for Human and Software Agents, in
Proceedings of the Second International Conference on Autonomous agents, May
Appendix A: List of Publications
Gobbin, R 1998, Adoption or Rejection: Information Systems and their Cultural
Fitness, Information Systems and Activity Theory: Tools in Context, in H Hasan, E
Gould, & P Hyland (Eds.), Wollongong University Press.
Gobbin, R 1998, The Role of Cultural Fitness in User Resistance to Information
Technology Tools, Interacting with Computers, 9, 275-285, Elsevier Science UK.
Gobbin, R & Jentzch, R 2003, A Cooperative Communicative Intelligent Agent
Model for E-Commerce, Managing E-commerce and Mobile Computing
Technologies, J Mariga, (ed.) IRM Press US.
Jentzsch, R, Mohammadian, M & Gobbin, R 2004, A Framework for using RFID in
a Hospital, IAWTC 2004 Conference Proceedings.
Gobbin, R, Jentzsch, R & Mohammadian, M 2004, Using Agent subjective
properties in modelling multiple agent communication activities, Wollongong
Conference on Human Computer Interaction, University of Wollongong.
Gobbin, R 2004, The Application of Intentional Subjective Properties and Mediated
Communication Tools to Software Agents in Online Dispute Resolution
Environments, ISCAR 2004, University of Wollongong, Australasian Journal of
Information Sciences.
Gobbin, R & Mohammadian, M 2004, Intentional Subjective Properties as
Communication Modelling Tools for Online Dispute Resolution Software Agents,
IAWTIC 2004 Conference Proceedings, Brisbane, Australia.
Gobbin, R & Mohammadian, M 2005, The Role of Semantic Tools for the
Construction of Subjective Meaning in Multi-agents Communication Systems,
IEEE- IAWTIC 2005 Conference, Vienna,
Appendix B: Protégé Ontology Editor
Protégé Ontology Editor
Protégé is the result of various artificial intelligence (AI) and knowledge-modelling
projects from the Medical Informatics group at Stanford University. Protégé is a free,
open-source platform that provides a growing user community with a suite of tools to
construct domain models and knowledge-based applications with ontologies. At its core,
Protégé implements a rich set of knowledge-modelling structures and actions that
support the creation, visualization, and manipulation of ontologies in various
representation formats. Protégé can be customized to provide domain-friendly support
for creating knowledge models and entering data. Further, Protégé can be extended by
way of a plug-in architecture and a Java-based Application Programming Interface (API)
for building knowledge-based tools and applications.
Ontology describes the concepts and relationships that are important in a particular
domain, providing a vocabulary for that domain as well as a computerized specification
of the meaning of terms used in the vocabulary. Ontologies range from taxonomies and
classifications, database schemas, to fully axiomatized theories. In recent years, ontologies
have been adopted in many business and scientific communities as a way to share, reuse
and process domain knowledge. Ontologies are now central to many applications such as
scientific knowledge portals, information management and integration systems, electronic
commerce, and semantic web services.
The Protégé platform supports two main ways of modelling ontologies:
* The Protégé-Frames editor enables users to build and populate ontologies that are
frame-based, in accordance with the Open Knowledge Base Connectivity protocol
(OKBC). In this model, the ontology consists of a set of classes organized in a
subsumption hierarchy to represent a domain's salient concepts, a set of slots associated
to classes to describe their properties and relationships, and a set of instances of those
classes - individual exemplars of the concepts that hold specific values for their
* The Protégé-OWL editor shown in Fig. 46 and Bean Generator plug-in in Fig. 47
enables users to build ontologies for the Semantic Web, in particular in the W3C's Web
Ontology Language (OWL). “OWL ontology may include descriptions of classes,
properties and their instances. Given such ontology, the OWL formal semantics specifies
how to derive its logical consequences, i.e. facts not literally present in the ontology, but
entailed by the semantics. These entailments may be based on a single document or
multiple distributed documents that have been combined using defined OWL
mechanisms" (see the OWL Web Ontology Language Guide).
From a programmer's perspective, one of Protégé's most attractive features is that it
provides an open source API to plug in your own Java components and access the
domain models from your application. As a result, you can develop systems very rapidly:
just start with the underlying domain model, let Protégé generate the basic user interface,
and then gradually write widgets and plug-ins to customize look-and-feel and behaviour.
You can even give Protégé to your customers and, with little training, let them build their
own knowledge and requirement models.
Java Bean Generator Protégé Plug-in v 3.2.1
Appendix C: JADE Agent Development
JADE Agent Development Environment
JADE (Java Agent Development Framework) is a software Framework fully
implemented in Java language. It simplifies the implementation of multi-agent systems
through a middle-ware that complies with the FIPA specifications and through a set of
graphical tools that supports the debugging and deployment phases. The agent platform
can be distributed across machines (which not even need to share the same OS) and the
configuration can be controlled via a remote GUI. The configuration can be even
changed at run-time by moving agents from one machine to another one, as and when
required. JADE is completely implemented in Java language and the minimal system
requirement is the version 1.4 of JAVA (the run time environment or the JDK).
The synergy between the JADE platform and the LEAP libraries allows obtaining a
FIPA-compliant agent platform with reduced footprint and compatibility with mobile
Java environments down to J2ME-CLDC MIDP 1.0. The LEAP libraries have been
developed with the collaboration of the LEAP project and can be downloaded as an addon of JADE from this same Web site.
JADE is free software and is distributed by Telecom Italy, the copyright holder, in open
source software under the terms of the LGPL (Lesser General Public License Version 2).
Since May 2003, a JADE Board has been created for supervising the management of the
JADE Research Project. Currently the JADE Board lists 5members: Telecom Italy,
Motorola, Whitestein AG, Profactor GmbH, and France Telecom R&D.
The latest version of JADE that will be used for this research is JADE 3.7 released
on 2nd July 2009.
The goal of JADE is to simplify the development of multi-agent systems while ensuring
standard compliance through a comprehensive set of system services and agents in
compliance with the FIPA specifications: naming service and yellow-page service,
message transport and parsing service, and a library of FIPA interaction protocols ready
to be used.
The JADE Agent Platform complies with FIPA specifications and includes all those
mandatory components that manage the platform that is the ACC, the AMS, and the DF.
All agent communication is performed through message passing, where FIPA ACL is the
language to represent messages.
The agent platform can be distributed on several hosts. Only one Java application, and
therefore only one Java Virtual Machine (JVM), is executed on each host. Each JVM is
basically a container of agents that provides a complete run time environment for agent
execution and allows several agents to concurrently execute on the same host.
The communication architecture offers flexible and efficient messaging, where JADE
creates and manages a queue of incoming ACL messages, private to each agent; agents
can access their queue via a combination of several modes: blocking, polling, timeout and
pattern matching based. The full FIPA communication model has been implemented and
its components have been clearly distinct and fully integrated: interaction protocols,
envelope, ACL, content languages, encoding schemes, ontologies and, finally, transport
protocols. The transport mechanism, in particular, is like a chameleon because it adapts
to each situation, by transparently choosing the best available protocol. Java RMI, eventnotification, HTTP, and IIOP are currently used, but more protocols can be easily added
via the JADE interfaces. Most of the interaction protocols defined by FIPA are already
available and can be instantiated after defining the application-dependent behaviour of
each state of the protocol. SL and agent management ontology have been implemented
already, as well as the support for user-defined content languages and ontologies that can
be implemented, registered with agents, and automatically used by the framework.
Agents are implemented as one thread per agent, but agents often need to execute parallel
tasks. Further to the multi-thread solution, offered directly by the JAVA language, JADE
supports also scheduling of cooperative behaviours, where JADE schedules these tasks in
a light and effective way. The run-time includes also some ready to use behaviours for the
most common tasks in agent programming, such as FIPA interaction protocols, waking
under a certain condition, and structuring complex tasks as aggregations of simpler ones.
Among the others, JADE offers also a so-called JessBehaviour that allows full integration
with JESS, where JADE provides the shell of the agent and guarantees (where possible)
the FIPA compliance, while JESS is the engine of the agent that performs all the
necessary reasoning. One of the examples shows integration between JADE Jess, and
Protege. Jess will not be used in this research however it can be added on in further
research where an expert system interface with agents will be integrated with ALMA
The agent platform provides a Graphical User Interface (GUI) for the remote
management, monitoring and controlling of the status of agents, allowing, for example, to
stop and restart agents. The GUI allows also creating and starting the execution of an
agent on a remote host, provided that an agent container is already running. The GUI
allows also controlling other remote FIPA-compliant agent platforms.
A GUI of the DF can be launched from the Tools menu of the RMA. By using this GUI,
the user can interact with the DF: view the descriptions of the registered agents, register
and deregister agents, modify the description of registered agent, and also search for
agent descriptions. The GUI allows also federating the DF with other DF’s and creating a
complex network of domains and sub-domains of yellow pages. Any federated DF, even
if resident on a remote non-JADE agent platform can also be controlled by the same
GUI and the same basic operations (view / register / deregister / modify / search) can
be executed on the remote DF.
Upon the core of JADE a number of graphical tools have been implemented that
supports the debugging phase, usually quite complex in distributed systems.
The Dummy Agent is a simple yet very useful tool for inspecting message exchanges
among agents. The dummy agent facilitates validation of an agent interface before
integration into the MAS and facilitates interrogative testing in the event that an agent is
failing. The graphical interface provides support to edit, compose and send ACL
messages to agents, to receive and view messages from agents, and, eventually, to save/
load messages to/from disk. (Source: )
The Sniffer Agent, as the name itself points out, allows to tracks messages exchanged in a
JADE agent platform. When the user decides to sniff an agent, or a group of agents,
every message directed to or coming from that agent or group of agents is tracked and
displayed in the sniffer window. The user can view, save, and load, every message track
for later analysis.
The Introspector Agent allows the monitoring and control of the life-cycle of a running
agent and its exchanged messages, both the queue of sent and received messages.
The development of JADE is still continuing. Further improvements, enhancements, and
implementations have already been planned, most of them in collaboration with
interested users of the JADE community. (Source: )
Appendix D: LKIF JADE ALMA Classes
The original LKIF in OWL format was developed by the Leibniz Centre at the University
of Amsterdam Under EU FP6 program IST – 027655 (Breuker, Valente and Winkels
2004) I have modified the original LKIF OWL ontology for the ALMA experiments that
is using a JAVA Bean class system.
The LKIF JADE Classes after conversion by Protégé Bean Generator 3.2.1 is shown
JADE ALMA Legal Ontology conversion text:
// file: generated by ontology bean generator. DO NOT EDIT,
package Alma_Legev2.onto;
import jade.content.onto.*;
import jade.content.schema.*;
import jade.util.leap.HashMap;
import jade.content.lang.Codec;
import jade.core.CaseInsensitiveString;
/** file:
* @author ontology bean generator
* @version 2010/03/22, 13:00:55
public class ALMA_LegalOntology extends jade.content.onto.Ontology {
public static final String ONTOLOGY_NAME = "ALMA_Legal";
// The singleton instance of this ontology
private static ReflectiveIntrospector introspect = new ReflectiveIntrospector();
private static Ontology theInstance = new ALMA_LegalOntology();
public static Ontology getInstance() {
return theInstance;
public static final String SUBJECTIVE1="Subjective1";
public static final String ALMA="ALMA";
public static final String OBJECTIVE1="Objective1";
public static final String
public static final String
public static final String
public static final String
public static final String
public static final String
public static final String
public static final String MODIFICATION_SUSPENSION="Modification_Suspension";
public static final String
public static final String
public static final String
public static final String
public static final String
public static final String
public static final String
public static final String LACTION_MANDATE="Laction_Mandate";
public static final String MODIFICATION_RELOCATION="Modification_Relocation";
public static final String
public static final String
public static final String MODIFICATION_REPEAL="Modification_Repeal";
public static final String
public static final String
public static final String
public static final String LACTION_ASSIGNMENT="Laction_Assignment";
public static final String PROCESS_CONTINUATION="Process_Continuation";
public static final String LACTION_DELEGATION="Laction_Delegation";
public static final String LACTION_ACT_OF_LAW="Laction_Act_of_Law";
public static final String ACTION_ORGANISATION="Action_Organisation";
public static final String LACTION_LEGAL_PERSON="Laction_Legal_Person";
public static final String PROCESS_INITIATION="Process_Initiation";
public static final String PROCESS_TERMINATION="Process_Termination";
public static final String LACTION_CORPORATION="Laction_Corporation";
public static final String LACTION_FOUNDATION="Laction_Foundation";
public static final String
public static final String LACTION_COMPANY="Laction_Company";
public static final String TOP_PHYSICAL_CONCEPT="Top_Physical_Concept";
public static final String PROCESS_PHYSICAL_OBJECT="Process_Physical_Object";
public static final String LACTION_PUBLIC_BODY="Laction_Public_Body";
public static final String
public static final String
public static final String EXPRESSION_LIE="Expression_Lie";
public static final String ACTION_PLAN="Action_Plan";
public static final String EXPRESSION_DESIRE="Expression_Desire";
public static final String ACTION_TRANSACTION="Action_Transaction";
public static final String
public static final String ACTION_PERSONAL_PLAN="Action_Personal_Plan";
public static final String ACTION_TRADE="Action_Trade";
public static final String PROCESS_PROCESS="Process_Process";
public static final String
public static final String PROCESS_CHANGE="Process_Change";
public static final String EXPRESSION_SPEECH_ACT="Expression_Speech_Act";
public static final String ACTION_CREATION="Action_Creation";
public static final String LACTION_DECISION="Laction_Decision";
public static final String
public static final String LACTION_PUBLIC_ACT="Laction_Public_Act";
public static final String ACTION_REACTION="Action_Reaction";
public static final String
public static final String
public static final String
public static final String
public static final String
public static final String
public static final String
public static final String MODIFICATION_REMAKING="Modification_Remaking";
public static final String
public static final String
public static final String
public static final String
public static final String
public static final String
public static final String MODIFICATION_VARIATION="Modification_Variation";
public static final String
public static final String MODIFICATION_EXCEPTION="Modification_Exception";
public static final String MODIFICATION_EXTENSION="Modification_Extension";
public static final String
public static final String ROLE_ORGANISATION_ROLE="Role_Organisation_Role";
public static final String ROLE_SOCIAL_ROLE="Role_Social_Role";
public static final String ROLE_ROLE_ROLE_PLAYED_BY="role_played_by";
public static final String ROLE_ROLE="Role_Role";
public static final String TOP_MENTAL_CONCEPT="Top_Mental_Concept";
public static final String
public static final String
public static final String EXPRESSION_QUALIFIED="Expression_Qualified";
public static final String EXPRESSION_ARGUMENT="Expression_Argument";
public static final String EXPRESSION_REASON="Expression_Reason";
public static final String ROLE_EPISTEMIC_ROLE="Role_Epistemic_Role";
public static final String ROLE_FUNCTION="Role_Function";
public static final String
public static final String LROLE_SOCIAL_LEGAL_ROLE="Lrole_Social_Legal_Role";
public static final String LROLE_LEGAL_ROLE="Lrole_Legal_Role";
public static final String ROLE_PERSON_ROLE="Role_Person_Role";
public static final String EXPRESSION_EVIDENCE="Expression_Evidence";
public static final String EXPRESSION_CAUSE="Expression_Cause";
public static final String EXPRESSION_FACT="Expression_Fact";
public static final String EXPRESSION_ASSUMPTION="Expression_Assumption";
public static final String EXPRESSION_SURPRISE="Expression_Surprise";
public static final String EXPRESSION_EXPECTATION="Expression_Expectation";
public static final String EXPRESSION_OBSERVATION="Expression_Observation";
public static final String EXPRESSION_PROBLEM="Expression_Problem";
public static final String RULES_ARGUMENT="Rules_Argument";
public static final String RULES_NEGATED_ATOM="Rules_Negated_Atom";
public static final String RULES_RULE="Rules_Rule";
public static final String RULES_VALID_RULE="Rules_Valid_Rule";
public static final String RULES_ASSUMPTION="Rules_Assumption";
public static final String RULES_EXCEPTION="Rules_Exception";
public static final String EXPRESSION_EXCEPTION="Expression_Exception";
public static final String RULES_ATOM="Rules_Atom";
public static final String
public static final String
public static final String
public static final String
public static final String
public static final String
public static final String PLACE_LOCATION_COMPLEX="Place_Location_Complex";
public static final String
public static final String MODIFICATION_ANNULMENT="Modification_Annulment";
public static final String
public static final String MODIFICATION_RENEWAL="Modification_Renewal";
public static final String
public static final String
public static final String EXPRESSION_DECLARATION="Expression_Declaration";
public static final String
public static final String
public static final String
public static final String
public static final String EXPRESSION_EXPRESSION="Expression_Expression";
public static final String
public static final String EXPRESSION_PROMISE="Expression_Promise";
public static final String
public static final String EXPRESSION_QUALIFICATION="Expression_Qualification";
public static final String
public static final String EXPRESSION_ASSERTION="Expression_Assertion";
public static final String
public static final String EXPRESSION_MEDIUM="Expression_Medium";
public static final String
public static final String
public static final String
public static final String
public static final String EXPRESSION_BELIEF="Expression_Belief";
public static final String
public static final String TOP_MENTAL_OBJECT="Top_Mental_Object";
public static final String
public static final String
public static final String
public static final String
public static final String
public static final String
public static final String
public static final String
public static final String ACTION_AGENT_ACTION_ACTOR_IN="action_actor_in";
public static final String
public static final String
public static final String ACTION_AGENT="Action_Agent";
public static final String
public static final String EXPRESSION_PROPOSITION="Expression_Proposition";
public static final String ACTION_ACTION_ACTION_ACTOR="action_actor";
public static final String ACTION_ACTION="Action_Action";
public static final String
public static final String EXPRESSION_INTENTION="Expression_Intention";
public static final String
public static final String
public static final String
public static final String MEREO_WHOLE="Mereo_Whole";
public static final String
public static final String TIME_PAIR_OF_PERIODS="Time_Pair_Of_Periods";
public static final String TOP_ABSTRACT_CONCEPT="Top_Abstract_Concept";
public static final String
public static final String
public static final String PLACE_PLACE_PLACE_IN="place_in";
public static final String
public static final String
public static final String
public static final String PLACE_PLACE_PLACE_MEET="place_meet";
public static final String PLACE_PLACE_PLACE_ABUT="place_abut";
public static final String
public static final String
public static final String PLACE_PLACE_PLACE_OVERLAP="place_overlap";
public static final String PLACE_PLACE_PLACE_CONNECT="place_connect";
public static final String
public static final String PLACE_PLACE_PLACE_COVER="place_cover";
public static final String PLACE_PLACE="Place_Place";
public static final String MEREO_ATOM="Mereo_Atom";
public static final String
public static final String
public static final String PLACE_RELATIVE_PLACE="Place_Relative_Place";
public static final String PLACE_ABSOLUTE_PLACE="Place_Absolute_Place";
public static final String
public static final String ACTION_NATURAL_OBJECT="Action_Natural_Object";
public static final String ACTION_ARTIFACT="Action_Artifact";
public static final String TOP_OCCURRENCE="Top_Occurrence";
public static final String
public static final String
public static final String
public static final String
public static final String TIME_MOMENT="Time_Moment";
public static final String
public static final String
public static final String
public static final String TIME_INTERVAL="Time_Interval";
public static final String EXPRESSION_DOCUMENT="Expression_Document";
public static final String MEREO_PAIR="Mereo_Pair";
public static final String
public static final String MEREO_COMPOSITION="Mereo_Composition";
public static final String MEREO_PART="Mereo_Part";
public static final String
public static final String
public static final String
public static final String
public static final String
public static final String
public static final String
public static final String
* Constructor
private ALMA_LegalOntology(){
super(ONTOLOGY_NAME, BasicOntology.getInstance());
try {
// adding Concept(s)
ConceptSchema modification_Application_IntervalSchema = new
ConceptSchema time_Temporal_OccurrenceSchema = new
ConceptSchema mereo_PartSchema = new ConceptSchema(MEREO_PART);
add(mereo_PartSchema, Alma_Legev2.onto.Mereo_Part.class);
ConceptSchema mereo_CompositionSchema = new
add(mereo_CompositionSchema, Alma_Legev2.onto.Mereo_Composition.class);
ConceptSchema modification_Efficacy_IntervalSchema = new
ConceptSchema mereo_PairSchema = new ConceptSchema(MEREO_PAIR);
add(mereo_PairSchema, Alma_Legev2.onto.Mereo_Pair.class);
ConceptSchema expression_DocumentSchema = new
add(expression_DocumentSchema, Alma_Legev2.onto.Expression_Document.class);
ConceptSchema time_IntervalSchema = new ConceptSchema(TIME_INTERVAL);
add(time_IntervalSchema, Alma_Legev2.onto.Time_Interval.class);
ConceptSchema modification_Application_DateSchema = new
ConceptSchema time_MomentSchema = new ConceptSchema(TIME_MOMENT);
add(time_MomentSchema, Alma_Legev2.onto.Time_Moment.class);
ConceptSchema top_Spatio_Temporal_OccurrenceSchema = new
ConceptSchema top_OccurrenceSchema = new ConceptSchema(TOP_OCCURRENCE);
add(top_OccurrenceSchema, Alma_Legev2.onto.Top_Occurrence.class);
ConceptSchema action_ArtifactSchema = new ConceptSchema(ACTION_ARTIFACT);
add(action_ArtifactSchema, Alma_Legev2.onto.Action_Artifact.class);
ConceptSchema action_Natural_ObjectSchema = new
add(action_Natural_ObjectSchema, Alma_Legev2.onto.Action_Natural_Object.class);
ConceptSchema modification_Dynamic_Temporal_EntitySchema = new
ConceptSchema place_Absolute_PlaceSchema = new
add(place_Absolute_PlaceSchema, Alma_Legev2.onto.Place_Absolute_Place.class);
ConceptSchema place_Relative_PlaceSchema = new
add(place_Relative_PlaceSchema, Alma_Legev2.onto.Place_Relative_Place.class);
ConceptSchema place_Comprehensive_PlaceSchema = new
ConceptSchema modification_Publication_DateSchema = new
ConceptSchema mereo_AtomSchema = new ConceptSchema(MEREO_ATOM);
add(mereo_AtomSchema, Alma_Legev2.onto.Mereo_Atom.class);
ConceptSchema place_PlaceSchema = new ConceptSchema(PLACE_PLACE);
add(place_PlaceSchema, Alma_Legev2.onto.Place_Place.class);
ConceptSchema top_Abstract_ConceptSchema = new
add(top_Abstract_ConceptSchema, Alma_Legev2.onto.Top_Abstract_Concept.class);
ConceptSchema time_Pair_Of_PeriodsSchema = new
add(time_Pair_Of_PeriodsSchema, Alma_Legev2.onto.Time_Pair_Of_Periods.class);
ConceptSchema modification_Static_Temporal_EntitySchema = new
ConceptSchema mereo_WholeSchema = new ConceptSchema(MEREO_WHOLE);
add(mereo_WholeSchema, Alma_Legev2.onto.Mereo_Whole.class);
ConceptSchema modification_Existence_DateSchema = new
ConceptSchema modification_Enter_in_Force_DateSchema = new
ConceptSchema modification_Delivery_DateSchema = new
ConceptSchema expression_IntentionSchema = new
add(expression_IntentionSchema, Alma_Legev2.onto.Expression_Intention.class);
ConceptSchema action_ActionSchema = new ConceptSchema(ACTION_ACTION);
add(action_ActionSchema, Alma_Legev2.onto.Action_Action.class);
ConceptSchema expression_PropositionSchema = new
add(expression_PropositionSchema, Alma_Legev2.onto.Expression_Proposition.class);
ConceptSchema action_AgentSchema = new ConceptSchema(ACTION_AGENT);
add(action_AgentSchema, Alma_Legev2.onto.Action_Agent.class);
ConceptSchema expression_Communicated_AttitudeSchema = new
ConceptSchema modification_In_Force_IntervalSchema = new
ConceptSchema top_Mental_ObjectSchema = new
add(top_Mental_ObjectSchema, Alma_Legev2.onto.Top_Mental_Object.class);
ConceptSchema expression_BeliefSchema = new
add(expression_BeliefSchema, Alma_Legev2.onto.Expression_Belief.class);
ConceptSchema expression_Propositional_AttitudeSchema = new
ConceptSchema expression_MediumSchema = new
add(expression_MediumSchema, Alma_Legev2.onto.Expression_Medium.class);
ConceptSchema expression_AssertionSchema = new
add(expression_AssertionSchema, Alma_Legev2.onto.Expression_Assertion.class);
ConceptSchema expression_QualificationSchema = new
ConceptSchema expression_PromiseSchema = new
add(expression_PromiseSchema, Alma_Legev2.onto.Expression_Promise.class);
ConceptSchema expression_ExpressionSchema = new
add(expression_ExpressionSchema, Alma_Legev2.onto.Expression_Expression.class);
ConceptSchema expression_DeclarationSchema = new
add(expression_DeclarationSchema, Alma_Legev2.onto.Expression_Declaration.class);
ConceptSchema modification_End_in_ForceSchema = new
ConceptSchema modification_RenewalSchema = new
add(modification_RenewalSchema, Alma_Legev2.onto.Modification_Renewal.class);
ConceptSchema modification_AnnulmentSchema = new
ConceptSchema place_Location_ComplexSchema = new
ConceptSchema expression_Evaluative_AttitudeSchema = new
ConceptSchema expression_Evaluative_PropositionSchema = new
ConceptSchema rules_AtomSchema = new ConceptSchema(RULES_ATOM);
add(rules_AtomSchema, Alma_Legev2.onto.Rules_Atom.class);
ConceptSchema expression_ExceptionSchema = new
add(expression_ExceptionSchema, Alma_Legev2.onto.Expression_Exception.class);
ConceptSchema rules_ExceptionSchema = new ConceptSchema(RULES_EXCEPTION);
add(rules_ExceptionSchema, Alma_Legev2.onto.Rules_Exception.class);
ConceptSchema rules_AssumptionSchema = new
add(rules_AssumptionSchema, Alma_Legev2.onto.Rules_Assumption.class);
ConceptSchema rules_Valid_RuleSchema = new
add(rules_Valid_RuleSchema, Alma_Legev2.onto.Rules_Valid_Rule.class);
ConceptSchema rules_RuleSchema = new ConceptSchema(RULES_RULE);
add(rules_RuleSchema, Alma_Legev2.onto.Rules_Rule.class);
ConceptSchema rules_Negated_AtomSchema = new
add(rules_Negated_AtomSchema, Alma_Legev2.onto.Rules_Negated_Atom.class);
ConceptSchema rules_ArgumentSchema = new
add(rules_ArgumentSchema, Alma_Legev2.onto.Rules_Argument.class);
ConceptSchema expression_ProblemSchema = new
add(expression_ProblemSchema, Alma_Legev2.onto.Expression_Problem.class);
ConceptSchema expression_ObservationSchema = new
add(expression_ObservationSchema, Alma_Legev2.onto.Expression_Observation.class);
ConceptSchema expression_ExpectationSchema = new
add(expression_ExpectationSchema, Alma_Legev2.onto.Expression_Expectation.class);
ConceptSchema expression_SurpriseSchema = new
add(expression_SurpriseSchema, Alma_Legev2.onto.Expression_Surprise.class);
ConceptSchema expression_AssumptionSchema = new
add(expression_AssumptionSchema, Alma_Legev2.onto.Expression_Assumption.class);
ConceptSchema expression_FactSchema = new ConceptSchema(EXPRESSION_FACT);
add(expression_FactSchema, Alma_Legev2.onto.Expression_Fact.class);
ConceptSchema expression_CauseSchema = new
add(expression_CauseSchema, Alma_Legev2.onto.Expression_Cause.class);
ConceptSchema expression_EvidenceSchema = new
add(expression_EvidenceSchema, Alma_Legev2.onto.Expression_Evidence.class);
ConceptSchema role_Person_RoleSchema = new
add(role_Person_RoleSchema, Alma_Legev2.onto.Role_Person_Role.class);
ConceptSchema lrole_Legal_RoleSchema = new
add(lrole_Legal_RoleSchema, Alma_Legev2.onto.Lrole_Legal_Role.class);
ConceptSchema lrole_Social_Legal_RoleSchema = new
ConceptSchema lrole_Professional_Legal_RoleSchema = new
ConceptSchema role_FunctionSchema = new ConceptSchema(ROLE_FUNCTION);
add(role_FunctionSchema, Alma_Legev2.onto.Role_Function.class);
ConceptSchema role_Epistemic_RoleSchema = new
add(role_Epistemic_RoleSchema, Alma_Legev2.onto.Role_Epistemic_Role.class);
ConceptSchema expression_ReasonSchema = new
add(expression_ReasonSchema, Alma_Legev2.onto.Expression_Reason.class);
ConceptSchema expression_ArgumentSchema = new
add(expression_ArgumentSchema, Alma_Legev2.onto.Expression_Argument.class);
ConceptSchema expression_QualifiedSchema = new
add(expression_QualifiedSchema, Alma_Legev2.onto.Expression_Qualified.class);
ConceptSchema top_Mental_ConceptSchema = new
add(top_Mental_ConceptSchema, Alma_Legev2.onto.Top_Mental_Concept.class);
ConceptSchema role_RoleSchema = new ConceptSchema(ROLE_ROLE);
add(role_RoleSchema, Alma_Legev2.onto.Role_Role.class);
ConceptSchema role_Social_RoleSchema = new
add(role_Social_RoleSchema, Alma_Legev2.onto.Role_Social_Role.class);
ConceptSchema role_Organisation_RoleSchema = new
add(role_Organisation_RoleSchema, Alma_Legev2.onto.Role_Organisation_Role.class);
ConceptSchema modification_Temporal_ModificationSchema = new
ConceptSchema modification_ExtensionSchema = new
add(modification_ExtensionSchema, Alma_Legev2.onto.Modification_Extension.class);
ConceptSchema modification_ExceptionSchema = new
add(modification_ExceptionSchema, Alma_Legev2.onto.Modification_Exception.class);
ConceptSchema modification_Modification_of_ScopeSchema = new
ConceptSchema modification_VariationSchema = new
add(modification_VariationSchema, Alma_Legev2.onto.Modification_Variation.class);
ConceptSchema modification_InterpretationSchema = new
ConceptSchema modification_Modification_of_TermSchema = new
ConceptSchema modification_Modification_of_MeaningSchema = new
ConceptSchema modification_DeregulationSchema = new
ConceptSchema modification_RatificationSchema = new
ConceptSchema modification_ApplicationSchema = new
ConceptSchema modification_RemakingSchema = new
add(modification_RemakingSchema, Alma_Legev2.onto.Modification_Remaking.class);
ConceptSchema modification_TranspositionSchema = new
ConceptSchema modification_Modification_of_SystemSchema = new
ConceptSchema modification_Semantic_AnnotationSchema = new
ConceptSchema modification_ModificationSchema = new
ConceptSchema action_ReactionSchema = new
add(action_ReactionSchema, Alma_Legev2.onto.Action_Reaction.class);
ConceptSchema laction_Public_ActSchema = new
add(laction_Public_ActSchema, Alma_Legev2.onto.Laction_Public_Act.class);
ConceptSchema laction_Legal_Speech_ActSchema = new
ConceptSchema laction_DecisionSchema = new
add(laction_DecisionSchema, Alma_Legev2.onto.Laction_Decision.class);
ConceptSchema action_CreationSchema = new ConceptSchema(ACTION_CREATION);
add(action_CreationSchema, Alma_Legev2.onto.Action_Creation.class);
ConceptSchema expression_Speech_ActSchema = new
add(expression_Speech_ActSchema, Alma_Legev2.onto.Expression_Speech_Act.class);
ConceptSchema process_ChangeSchema = new ConceptSchema(PROCESS_CHANGE);
add(process_ChangeSchema, Alma_Legev2.onto.Process_Change.class);
ConceptSchema process_ProcessSchema = new ConceptSchema(PROCESS_PROCESS);
add(process_ProcessSchema, Alma_Legev2.onto.Process_Process.class);
ConceptSchema action_TradeSchema = new ConceptSchema(ACTION_TRADE);
add(action_TradeSchema, Alma_Legev2.onto.Action_Trade.class);
ConceptSchema action_Personal_PlanSchema = new
add(action_Personal_PlanSchema, Alma_Legev2.onto.Action_Personal_Plan.class);
ConceptSchema action_Collaborative_PlanSchema = new
ConceptSchema action_TransactionSchema = new
add(action_TransactionSchema, Alma_Legev2.onto.Action_Transaction.class);
ConceptSchema expression_DesireSchema = new
add(expression_DesireSchema, Alma_Legev2.onto.Expression_Desire.class);
ConceptSchema action_PlanSchema = new ConceptSchema(ACTION_PLAN);
add(action_PlanSchema, Alma_Legev2.onto.Action_Plan.class);
ConceptSchema expression_LieSchema = new ConceptSchema(EXPRESSION_LIE);
add(expression_LieSchema, Alma_Legev2.onto.Expression_Lie.class);
ConceptSchema expression_Statement_In_WritingSchema = new
ConceptSchema laction_Legislative_BodySchema = new
ConceptSchema laction_Public_BodySchema = new
add(laction_Public_BodySchema, Alma_Legev2.onto.Laction_Public_Body.class);
ConceptSchema process_Physical_ObjectSchema = new
ConceptSchema top_Physical_ConceptSchema = new
add(top_Physical_ConceptSchema, Alma_Legev2.onto.Top_Physical_Concept.class);
ConceptSchema laction_CompanySchema = new
add(laction_CompanySchema, Alma_Legev2.onto.Laction_Company.class);
ConceptSchema laction_Private_Legal_PersonSchema = new
ConceptSchema laction_FoundationSchema = new
add(laction_FoundationSchema, Alma_Legev2.onto.Laction_Foundation.class);
ConceptSchema laction_CorporationSchema = new
add(laction_CorporationSchema, Alma_Legev2.onto.Laction_Corporation.class);
ConceptSchema process_TerminationSchema = new
add(process_TerminationSchema, Alma_Legev2.onto.Process_Termination.class);
ConceptSchema process_InitiationSchema = new
add(process_InitiationSchema, Alma_Legev2.onto.Process_Initiation.class);
ConceptSchema laction_Legal_PersonSchema = new
add(laction_Legal_PersonSchema, Alma_Legev2.onto.Laction_Legal_Person.class);
ConceptSchema action_OrganisationSchema = new
add(action_OrganisationSchema, Alma_Legev2.onto.Action_Organisation.class);
ConceptSchema laction_Act_of_LawSchema = new
add(laction_Act_of_LawSchema, Alma_Legev2.onto.Laction_Act_of_Law.class);
ConceptSchema laction_DelegationSchema = new
add(laction_DelegationSchema, Alma_Legev2.onto.Laction_Delegation.class);
ConceptSchema process_ContinuationSchema = new
add(process_ContinuationSchema, Alma_Legev2.onto.Process_Continuation.class);
ConceptSchema laction_AssignmentSchema = new
add(laction_AssignmentSchema, Alma_Legev2.onto.Laction_Assignment.class);
ConceptSchema modification_IntegrationSchema = new
ConceptSchema modification_RepealSchema = new
add(modification_RepealSchema, Alma_Legev2.onto.Modification_Repeal.class);
ConceptSchema modification_RelocationSchema = new
ConceptSchema laction_MandateSchema = new
add(laction_MandateSchema, Alma_Legev2.onto.Laction_Mandate.class);
ConceptSchema modification_Prorogation_in_ForceSchema = new
ConceptSchema modification_Start_in_ForceSchema = new
ConceptSchema modification_Textual_ModificationSchema = new
ConceptSchema modification_SubstitutionSchema = new
ConceptSchema modification_Start_EfficacySchema = new
ConceptSchema modification_SuspensionSchema = new
ConceptSchema modification_Prorogation_EfficacySchema = new
ConceptSchema modification_In_Force_ModificationSchema = new
ConceptSchema modification_Efficacy_ModificationSchema = new
ConceptSchema modification_UltractivitySchema = new
ConceptSchema modification_End_efficacySchema = new
ConceptSchema modification_RetroactivitySchema = new
// adding AgentAction(s)
// adding AID(s)
ConceptSchema objective1Schema = new ConceptSchema(OBJECTIVE1);
add(objective1Schema, Alma_Legev2.onto.Objective1.class);
ConceptSchema almaSchema = new ConceptSchema(ALMA);
add(almaSchema, Alma_Legev2.onto.ALMA.class);
ConceptSchema subjective1Schema = new ConceptSchema(SUBJECTIVE1);
add(subjective1Schema, Alma_Legev2.onto.Subjective1.class);
// adding Predicate(s)
// adding fields
MEDIATLY_AFTER, time_Temporal_OccurrenceSchema, 0,
TER, time_Temporal_OccurrenceSchema, 0, ObjectSchema.UNLIMITED);
FORE, time_Temporal_OccurrenceSchema, 0, ObjectSchema.UNLIMITED);
MPORAL_RELATION, time_Temporal_OccurrenceSchema, 0,
MEDIATLY_BEFORE, time_Temporal_OccurrenceSchema, 0,
TWEEN, time_Pair_Of_PeriodsSchema, 0, ObjectSchema.UNLIMITED);
time_MomentSchema, 0, ObjectSchema.UNLIMITED);
time_MomentSchema, 0, ObjectSchema.UNLIMITED);
time_IntervalSchema, 0, ObjectSchema.UNLIMITED);
(TermSchema)getSchema(BasicOntology.STRING), 0, ObjectSchema.UNLIMITED);
time_IntervalSchema, 0, ObjectSchema.UNLIMITED);
place_PlaceSchema.add(PLACE_PLACE_PLACE_COVER, place_PlaceSchema, 0,
place_PlaceSchema, 0, ObjectSchema.UNLIMITED);
place_PlaceSchema.add(PLACE_PLACE_PLACE_CONNECT, place_PlaceSchema, 0,
place_PlaceSchema.add(PLACE_PLACE_PLACE_OVERLAP, place_PlaceSchema, 0,
place_Location_ComplexSchema, ObjectSchema.OPTIONAL);
place_PlaceSchema, 0, ObjectSchema.UNLIMITED);
place_PlaceSchema.add(PLACE_PLACE_PLACE_ABUT, place_PlaceSchema, 0,
place_PlaceSchema.add(PLACE_PLACE_PLACE_MEET, place_PlaceSchema, 0,
place_PlaceSchema, 0, ObjectSchema.UNLIMITED);
place_PlaceSchema, 0, ObjectSchema.UNLIMITED);
place_PlaceSchema.add(PLACE_PLACE_PLACE_IN, new ConceptSchema("Concept"),
0, ObjectSchema.UNLIMITED);
place_PlaceSchema, 0, ObjectSchema.UNLIMITED);
place_PlaceSchema, 0, ObjectSchema.UNLIMITED);
D_BY, action_AgentSchema, 0, ObjectSchema.UNLIMITED);
action_AgentSchema, 0, ObjectSchema.UNLIMITED);
TUDE, expression_Propositional_AttitudeSchema, 0, ObjectSchema.UNLIMITED);
expression_BeliefSchema, 0, ObjectSchema.UNLIMITED);
expression_Communicated_AttitudeSchema, 0, ObjectSchema.UNLIMITED);
action_ActionSchema, 0, ObjectSchema.UNLIMITED);
top_Mental_ObjectSchema, 0, ObjectSchema.UNLIMITED);
expression_BeliefSchema, 0, ObjectSchema.UNLIMITED);
expression_IntentionSchema, 0, ObjectSchema.UNLIMITED);
TITUDE_EXPRESSION_STATES, expression_PropositionSchema, 0,
action_AgentSchema, 0, ObjectSchema.UNLIMITED);
action_AgentSchema, 0, ObjectSchema.UNLIMITED);
action_AgentSchema, 0, ObjectSchema.UNLIMITED);
UDE_EXPRESSION_TOWARDS, expression_PropositionSchema, 0,
expression_ExpressionSchema, 0, ObjectSchema.UNLIMITED);
S, expression_ExpressionSchema, 0, ObjectSchema.UNLIMITED);
QUALIFIES, expression_QualifiedSchema, 0, ObjectSchema.UNLIMITED);
expression_ExpressionSchema, 0, ObjectSchema.UNLIMITED);
RTED_BY, expression_AssertionSchema, 0, ObjectSchema.UNLIMITED);
ISED_BY, expression_PromiseSchema, 0, ObjectSchema.UNLIMITED);
ARED_BY, expression_DeclarationSchema, 0, ObjectSchema.UNLIMITED);
UM, expression_MediumSchema, 0, ObjectSchema.UNLIMITED);
CLARES, expression_ExpressionSchema, 0, ObjectSchema.UNLIMITED);
ODUCE_TEXTUAL_MODIFICATION, modification_End_in_ForceSchema,
N_PRODUCE_TEXTUAL_MODIFICATION, modification_End_in_ForceSchema,
ION_COMPLEX_FOR, place_PlaceSchema, 0, ObjectSchema.UNLIMITED);
EXPRESSION_EVALUATES, expression_Evaluative_PropositionSchema, 0,
expression_Evaluative_PropositionSchema, 0, ObjectSchema.UNLIMITED);
TION_EXPRESSION_EVALUATED_BY, expression_Evaluative_AttitudeSchema, 0,
ATIVELY_COMPARABLE, expression_QualifiedSchema, 0,
ED_BY, expression_QualificationSchema, 0, ObjectSchema.UNLIMITED);
TION_EFFICACY, modification_Efficacy_IntervalSchema, 0,
TION_IN_FORCE, modification_In_Force_IntervalSchema, 0,
TION_APPLICATION, modification_Application_DateSchema, 0,
N_PRODUCE_EFFICACY_MODIFICATION, new ConceptSchema("Concept"), 0,
N_PRODUCE_INFORCE_MODIFICATION, new ConceptSchema("Concept"), 0,
CE_EFFICACY_MODIFICATION, new ConceptSchema("Concept"), 0,
CE_INFORCE_MODIFICATION, new ConceptSchema("Concept"), 0,
ON_PRODUCE_EFFICACY_MODIFICATION, new ConceptSchema("Concept"), 0,
ON_PRODUCE_INFORCE_MODIFICATION, new ConceptSchema("Concept"), 0,
_DURATION, time_IntervalSchema, 0, ObjectSchema.UNLIMITED);
// adding name mappings
// adding inheritance
}catch (java.lang.Exception e) {e.printStackTrace();}
Appendix E: Eclipse Java IDE
Eclipse Java IDE
The Java IDE Eclipse from the Apache software foundation used in the research
was chosen for its Gnu Licence and its wide use and support. Another important
factor was the number of plug-ins available and able to be used immediately with
few modifications. The availability of the JADE Development plug-in EJade
provided by the University of Trento was another important determinant.
Details of Eclipse platform:
Eclipse IDE for Java Developers
Build id: 20100218-1602
(c) Copyright Eclipse contributors and others 2000, 2009. All rights reserved.
This product includes software developed by the
Apache Software Foundation
*** Date: Friday, 26 March 2010 9:41:50 AM AET
*** Platform Details:
*** System properties:
C:\Program Files\Java\jre6\bin\client\jvm.dll
[email protected]/../p2/
eclipse.vm=C:\Program Files\Java\jre6\bin\client\jvm.dll
java.endorsed.dirs=C:\Program Files\Java\jre6\lib\endorsed
java.ext.dirs=C:\Program Files\Java\jre6\lib\ext;C:\WINDOWS\Sun\Java\lib\ext
java.home=C:\Program Files\Java\jre6\DOCUME~1\s614023\LOCALS~1\Temp\
2;C:\WINDOWS;C:/Program Files/Java/jre6/bin/client;C:/Program
32\Wbem;C:\Program Files\QuickTime\QTSystem\;C:\Program
Files\Graphviz2.26.3\bin SE Runtime Environment
java.runtime.version=1.6.0_18-b07 Platform API Specification
java.specification.vendor=Sun Microsystems Inc.
java.vendor=Sun Microsystems Inc.
184 mode HotSpot(TM) Client VM Virtual Machine Specification
java.vm.specification.vendor=Sun Microsystems Inc.
java.vm.vendor=Sun Microsystems Inc.
os.arch=x86 XP
[email protected]:start
sun.boot.class.path=C:\Program Files\Java\jre6\lib\resources.jar;C:\Program
Files\Java\jre6\lib\rt.jar;C:\Program Files\Java\jre6\lib\sunrsasign.jar;C:\Program
Files\Java\jre6\lib\jsse.jar;C:\Program Files\Java\jre6\lib\jce.jar;C:\Program
Files\Java\jre6\lib\charsets.jar;C:\Program Files\Java\jre6\classes
sun.boot.library.path=C:\Program Files\Java\jre6\bin
sun.cpu.isalist=pentium_pro+mmx pentium_pro pentium+mmx pentium i486 i386
sun.jnu.encoding=Cp1252 Client Compiler
sun.os.patch.level=Service Pack 2
*** Features:
org.eclipse.cvs (1.1.101.R35x_v20100125-7E79FGD9kKF67BPDZ7PKCE1911)
"Eclipse CVS Client" ( "EPP Java Package" (1.1.2.R35x_v20100119-7e7eFAnFEx2X_fZltJhDjz-D) "Help
System Base"
org.eclipse.jdt (3.5.2.r352_v20100108-7r88FEwFI0WTuoBl0iaG0tyhfZH6)
"Eclipse Java Development Tools"
org.eclipse.mylyn_feature (3.2.3.v20100217-0100-e3x) "Mylyn"
org.eclipse.mylyn.bugzilla_feature (3.2.3.v20100217-0100-e3x) "Mylyn"
org.eclipse.mylyn.context_feature (3.2.3.v20100217-0100-e3x) "Mylyn"
org.eclipse.mylyn.ide_feature (3.2.3.v20100217-0100-e3x) "Mylyn"
org.eclipse.mylyn.java_feature (3.2.3.v20100217-0100-e3x) "Mylyn"
org.eclipse.mylyn.team_feature (3.2.3.v20100217-0100-e3x) "Mylyn"
org.eclipse.mylyn.wikitext_feature (1.1.3.v20100217-0100-e3x) "Mylyn"
org.eclipse.platform (3.5.2.R35x_v20100210-0800-9hEiFzmFstTiEn9hNYgDWg1XN8ulH_JvCNGB) "Eclipse Platform"
org.eclipse.rcp (3.5.2.R35x_v20100119-9SA0FxwFnoCU5XxWItFdXXb27BA6)
"Eclipse RCP"
org.eclipse.wst.xml_ui.feature (3.1.1.v2009071610317H6FMbDxtkMs9OeLGF98LRhdPKeo) "Eclipse XML Editors and Tools"
*** Plug-in Registry: (4.0.1.v20090822) "International Components for Unicode for Java
(ICU4J)" [Active]
com.jcraft.jsch (0.1.41.v200903070017) "JSch" [Resolved]
it.fbk.sra.ejade (0.8.0) "Ejade Plug-in" [Active]
javax.servlet (2.5.0.v200806031605) "Servlet API Bundle" [Resolved]
javax.servlet.jsp (2.0.0.v200806031607) "Java Server Pages API Bundle"
javax.xml (1.3.4.v200902170245) "JAXP XML" [Resolved]
org.apache.ant (1.7.1.v20090120-1145) "Apache Ant" [Resolved]
org.apache.commons.codec (1.3.0.v20080530-1600) "Apache Commons Codec
Plug-in" [Resolved]
org.apache.commons.codec (1.3.0.v20100106-1700) "Apache Commons Codec
Plug-in" [Resolved]
org.apache.commons.el (1.0.0.v200806031608) "Apache Commons JSP 2.0
Expression Language Interpreter" [Resolved]
org.apache.commons.httpclient (3.1.0.v20080605-1935) "Apache Commons
Httpclient" [Resolved]
org.apache.commons.lang (2.3.0.v200803061910) "Apache Jakarta Commons Lang"
org.apache.commons.logging (1.0.4.v200904062259) "Apache Commons Logging
Plug-in" [Resolved]
org.apache.jasper (5.5.17.v200903231320) "Apache Jasper 2 Plug-in" [Resolved]
org.apache.lucene (1.9.1.v20080530-1600) "Apache Lucene" [Resolved]
org.apache.lucene.analysis (1.9.1.v20080530-1600) "Apache Lucene Analysis"
org.apache.xerces (2.9.0.v200909240008) "Apache Xerces-J" [Resolved]
org.apache.xml.resolver (1.2.0.v200902170519) "Apache XmlResolver" [Resolved]
org.apache.xml.serializer (2.7.1.v200902170519) "Apache XML Commons
Serializer" [Resolved]
org.eclipse.ant.core (3.2.101.v20091110_r352) "Ant Build Tool Core" [Starting]
org.eclipse.ant.ui (3.4.2.v20091204_r352) "Ant UI" [Active] (3.5.2.r35x_20100113-0800) "Compare Support" [Starting] (3.5.0.I20090430-0408) "Core Compare Support" [Active] (1.0.0.I20090430-0408) "Compare Support for Word"
org.eclipse.core.boot (3.1.100.v20080218) "Core Boot" [Starting]
org.eclipse.core.commands (3.5.0.I20090525-2000) "Commands" [Resolved]
org.eclipse.core.contenttype (3.4.1.R35x_v20090826-0451) "Eclipse Content
Mechanism" [Active]
org.eclipse.core.databinding (1.2.0.M20090819-0800) "JFace Data Binding"
org.eclipse.core.databinding.beans (1.2.0.I20090525-2000) "JFace Data Binding for
JavaBeans" [Resolved]
org.eclipse.core.databinding.observable (1.2.0.M20090902-0800) "JFace Data
Binding Observables" [Active] (1.2.0.M20090819-0800) "JFace Data
Binding" [Starting]
org.eclipse.core.expressions (3.4.101.R35x_v20100209) "Expression Language"
org.eclipse.core.filebuffers (3.5.0.v20090526-2000) "File Buffers" [Active]
org.eclipse.core.filesystem (1.2.1.R35x_v20091203-1235) "Core File Systems"
org.eclipse.core.filesystem.win32.x86 (1.1.0.v20080604-1400) "Core File System
for Windows" [Resolved] (3.4.100.v20090429-1800) "Eclipse Jobs Mechanism" [Active] (1.2.1.r35x_20090812-1200) "Internet Connection
Management" [Active] (1.0.0.I20080909) "Proxy for Windows" [Resolved]
org.eclipse.core.resources (3.5.2.R35x_v20091203-1235) "Core Resource
Management" [Active]
org.eclipse.core.resources.compatibility (3.4.1.R35x_v20100113-0530) "Core
Resource Management Compatibility Fragment" [Resolved]
org.eclipse.core.resources.win32.x86 (3.5.0.v20081020) "Core Resource
Management Win32 Fragment" [Resolved]
org.eclipse.core.runtime (3.5.0.v20090525) "Core Runtime" [Active]
org.eclipse.core.runtime.compatibility (3.2.0.v20090413) "Core Runtime Plug-in
Compatibility" [Active]
org.eclipse.core.runtime.compatibility.auth (3.2.100.v20090413) "Authorization
Compatibility Plug-in" [Active]
org.eclipse.core.runtime.compatibility.registry (3.2.200.v20090429-1800) "Eclipse
Registry Compatibility Fragment" [Resolved]
org.eclipse.core.variables (3.2.200.v20090521) "Core Variables" [Active]
org.eclipse.cvs (1.0.400.v201002111343) "Eclipse CVS Client" [Starting]
org.eclipse.debug.core (3.5.1.v20091103_r352) "Debug Core" [Active]
org.eclipse.debug.ui (3.5.2.v20091028_r352) "Debug UI" [Active]
org.eclipse.draw2d (3.5.2.v20091126-1908) "Graphical Editing Framework
Draw2d" [Starting]
org.eclipse.ecf (3.0.0.v20090831-1906) "Eclipse Communication Framework
(ECF)" [Starting]
org.eclipse.ecf.filetransfer (3.0.0.v20090831-1906) "ECF Filetransfer API" [Active]
org.eclipse.ecf.identity (3.0.0.v20090831-1906) "ECF Identity API" [Starting]
org.eclipse.ecf.provider.filetransfer (3.0.1.v20090831-1906) "ECF Filetransfer
Provider" [Starting]
org.eclipse.ecf.provider.filetransfer.httpclient (3.0.1.v20090831-1906) "ECF
HttpClient Filetransfer Provider" [Starting]
org.eclipse.ecf.provider.filetransfer.httpclient.ssl (1.0.0.v20090831-1906) "ECF
HttpClient Filetransfer Provider" [Resolved]
org.eclipse.ecf.provider.filetransfer.ssl (1.0.0.v20090831-1906) "ECF Filetransfer
Provider" [Resolved]
org.eclipse.ecf.ssl (1.0.0.v20090831-1906) "Eclipse Communication Framework
(ECF)" [Resolved]
org.eclipse.emf.common (2.5.0.v200906151043) "EMF Common" [Starting]
org.eclipse.emf.common.ui (2.5.0.v200906151043) "EMF Common UI" [Starting]
org.eclipse.emf.ecore (2.5.0.v200906151043) "EMF Ecore" [Starting]
org.eclipse.emf.ecore.change (2.5.0.v200906151043) "EMF Change Model"
org.eclipse.emf.ecore.edit (2.5.0.v200906151043) "EMF Ecore Edit" [Starting]
org.eclipse.emf.ecore.xmi (2.5.0.v200906151043) "EMF XML/XMI Persistence"
org.eclipse.emf.edit (2.5.0.v200906151043) "EMF Edit" [Starting]
org.eclipse.emf.edit.ui (2.5.0.v200906151043) "EMF Edit UI" [Starting] ( "EPP Java Package"
org.eclipse.epp.usagedata.gathering (1.1.1.R201001291118) "Usage Data Gathering
Plug-in" [Active]
org.eclipse.epp.usagedata.recording (1.1.1.R201001291118) "Usage Data Recording
Plug-in" [Active]
org.eclipse.epp.usagedata.ui (1.1.1.R201001291118) "Usage Data UI Plug-in"
[Starting] (1.2.1.R35x_v20091203) "Equinox Application Container"
org.eclipse.equinox.common (3.5.1.R35x_v20090807-1100) "Common Eclipse
Runtime" [Active]
org.eclipse.equinox.concurrent (1.0.1.R35x_v20100209) "Equinox Concurrent API"
org.eclipse.equinox.ds (1.1.1.R35x_v20090806) "Declarative Services" [Active]
org.eclipse.equinox.frameworkadmin (1.0.100.v20090520-1905) "Equinox
Framework Admin" [Active]
org.eclipse.equinox.frameworkadmin.equinox (1.0.101.R35x_v20091214) "Equinox
Framework Admin for Equinox" [Active]
org.eclipse.equinox.http.jetty (2.0.0.v20090520-1800) "Jetty Http Service" [Starting]
org.eclipse.equinox.http.registry (1.0.200.v20090520-1800) "Http Service Registry
Extensions" [Resolved]
org.eclipse.equinox.http.servlet (1.0.200.v20090520-1800) "Http Services Servlet"
org.eclipse.equinox.jsp.jasper (1.0.200.v20090520-1800) "Jasper Jsp Support
Bundle" [Starting]
org.eclipse.equinox.jsp.jasper.registry (1.0.100.v20090520-1800) "Jasper Jsp
Registry Support Plug-in" [Starting]
org.eclipse.equinox.launcher (1.0.201.R35x_v20090715) "Equinox Launcher"
org.eclipse.equinox.launcher.win32.win32.x86 (1.0.200.v20090519) "Equinox
Launcher Win32 X86 Fragment" [Resolved]
org.eclipse.equinox.p2.artifact.repository (1.0.101.R35x_v20090721) "Equinox
Provisioning Artifact Repository Support" [Active]
org.eclipse.equinox.p2.console (1.0.100.v20090520-1905) "Equinox Provisioning
Console" [Starting]
org.eclipse.equinox.p2.core (1.0.101.R35x_v20090819) "Equinox Provisioning
Core" [Active]
org.eclipse.equinox.p2.director (1.0.101.R35x_v20100112) "Equinox Provisioning
Director" [Active] (1.0.101.R35x_v20091106) "Equinox
Provisioning Director Application" [Starting]
org.eclipse.equinox.p2.directorywatcher (1.0.100.v20090525) "Equinox
Provisioning Directory Watcher" [Active]
org.eclipse.equinox.p2.engine (1.0.102.R35x_v20091117) "Equinox Provisioning
Engine" [Active]
org.eclipse.equinox.p2.exemplarysetup (1.0.100.v20090520-1905) "Equinox
Provisioning Exemplary Setup" [Active]
org.eclipse.equinox.p2.extensionlocation (1.0.100.v20090520-1905) "Extension
Location Repository Support" [Active]
org.eclipse.equinox.p2.garbagecollector (1.0.100.v20090520-1905) "Provisioning
Garbage Collector" [Active]
org.eclipse.equinox.p2.jarprocessor (1.0.100.v20090520-1905) "Equinox
Provisioning JAR Processor" [Resolved]
org.eclipse.equinox.p2.metadata (1.0.101.R35x_v20100112) "Equinox Provisioning
Metadata" [Active]
org.eclipse.equinox.p2.metadata.generator (1.0.101.R35x_20100114) "Equinox
Provisioning Metadata Generator" [Starting]
org.eclipse.equinox.p2.metadata.repository (1.0.101.R35x_v20090812) "Equinox
Provisioning Metadata Repository" [Active]
org.eclipse.equinox.p2.publisher (1.0.1.R35x_20100105) "Equinox Provisioning
Publisher" [Active]
org.eclipse.equinox.p2.reconciler.dropins (1.0.100.v20090520-1905) "Dropin
Reconciler Plug-in" [Active]
org.eclipse.equinox.p2.repository (1.0.1.R35x_v20100105) "Equinox Provisioning
Repository" [Active] (1.0.2.R35x_20100111) "Equinox p2
repository tools." [Starting]
org.eclipse.equinox.p2.touchpoint.eclipse (1.0.101.R35x_20090820-1821) "Equinox
Provisioning Eclipse Touchpoint" [Active]
org.eclipse.equinox.p2.touchpoint.natives (1.0.101.R35x_v20090806) "Equinox
Provisioning Native Touchpoint" [Starting]
org.eclipse.equinox.p2.ui (1.0.101.R35x_v20090819) "Equinox Provisioning UI
Support" [Active]
org.eclipse.equinox.p2.ui.sdk (1.0.100.v20090520-1905) "Equinox Provisioning
Platform Update Support" [Active]
org.eclipse.equinox.p2.ui.sdk.scheduler (1.0.0.v20090520-1905) "Equinox
Provisioning Platform Automatic Update Support" [Active]
org.eclipse.equinox.p2.updatechecker (1.1.0.v20090520-1905) "Equinox
Provisioning Update Checker" [Active]
org.eclipse.equinox.p2.updatesite (1.0.101.R35x_20100105) "Update site repository
adapter bundle (Incubation)" [Starting]
org.eclipse.equinox.preferences (3.2.301.R35x_v20091117) "Eclipse Preferences
Mechanism" [Active]
org.eclipse.equinox.registry (3.4.100.v20090520-1800) "Extension Registry
Support" [Active] (1.0.100.v20090520-1800) "Equinox Java
Authentication and Authorization Service (JAAS)" [Active] (1.0.100.v20090520-1800) "Equinox Security
Default UI" [Starting] (1.0.100.v20090520-1800) "Windows Data
Protection services integration" [Resolved]
org.eclipse.equinox.simpleconfigurator (1.0.101.R35x_v20090807-1100) "Simple
Configurator" [Active]
org.eclipse.equinox.simpleconfigurator.manipulator (1.0.101.R35x_v20100209)
"Simple Configurator Manipulator" [Active]
org.eclipse.equinox.util (1.0.100.v20090520-1800) "Equinox Util Bundle" [Active]
org.eclipse.gef (3.5.1.v20090910-2020) "Graphical Editing Framework GEF"
[Starting] (3.4.1.v20090805_35x) "Help System Core" [Active] (3.1.400.v20090429_1800) "Help Application Server"
[Starting] (3.4.0.v201002111343) "Help System Base" [Active] (3.4.1.v20090819_35x) "Help System UI" [Active] (3.4.1.v20091009_35x) "Help System Webapp" [Starting]
org.eclipse.jdt (3.5.2.v201002111343) "Eclipse Java Development Tools"
org.eclipse.jdt.apt.core (3.3.202.R35x_v20091130-2300) "Java Annotation
Processing Core" [Active]
org.eclipse.jdt.apt.pluggable.core (1.0.201.R35x_v20090925-1100) "Java Compiler
Apt IDE" [Active]
org.eclipse.jdt.apt.ui (3.3.200.v20090930-2100_R35x) "Java Annotation Processing
UI" [Starting]
org.eclipse.jdt.compiler.apt (1.0.201.R35x_v20090925-1100) "Java Compiler Apt"
org.eclipse.jdt.compiler.tool (1.0.100.v_981_R35x) "Java Compiler Tool Support"
org.eclipse.jdt.core (3.5.2.v_981_R35x) "Java Development Tools Core" [Active]
org.eclipse.jdt.core.manipulation (1.3.0.v20090603) "Java Code Manipulation
Functionality" [Active]
org.eclipse.jdt.debug (3.5.0.v20090526) "JDI Debug Model" [Active]
org.eclipse.jdt.debug.ui (3.4.1.v20090811_r351) "JDI Debug UI" [Active]
org.eclipse.jdt.doc.user (3.5.2.r352_v20091015) "Eclipse Java development user
guide" [Resolved]
org.eclipse.jdt.junit (3.5.2.r352_v20100113-0800) "Java Development Tools JUnit
support" [Starting]
org.eclipse.jdt.junit.runtime (3.4.100.v20090513-2000) "Java Development Tools
JUnit runtime support" [Resolved]
org.eclipse.jdt.junit4.runtime (1.1.0.v20090513-2000) "Java Development Tools
JUnit4 runtime support" [Resolved]
org.eclipse.jdt.launching (3.5.1.v20100108_r352) "Java Development Tools
Launching Support" [Active]
org.eclipse.jdt.ui (3.5.2.r352_v20100106-0800) "Java Development Tools UI"
org.eclipse.jem.util (2.0.201.v201001252130) "Java EMF Model Utilities" [Starting]
org.eclipse.jface (3.5.2.M20100120-0800) "JFace" [Active]
org.eclipse.jface.databinding (1.3.1.M20090826-0800) "JFace Data Binding for
SWT and JFace" [Resolved]
org.eclipse.jface.text (3.5.2.r352_v20091118-0800) "JFace Text" [Resolved]
org.eclipse.jsch.core (1.1.100.I20090430-0408) "JSch Core" [Active]
org.eclipse.jsch.ui (1.1.200.r35x_20100210-1114) "JSch UI" [Starting]
org.eclipse.ltk.core.refactoring (3.5.0.v20090513-2000) "Refactoring Core" [Active]
org.eclipse.ltk.ui.refactoring (3.4.101.r352_v20100209) "Refactoring UI" [Active]
org.eclipse.mylyn (3.2.3.v20100217-0100-e3x) "Mylyn" [Resolved]
org.eclipse.mylyn.bugzilla.core (3.2.3.v20100217-0100-e3x) "Mylyn Bugzilla
Connector Core" [Active]
org.eclipse.mylyn.bugzilla.ide (3.2.3.v20100217-0100-e3x) "Mylyn Bugzilla IDE
Extensions" [Resolved]
org.eclipse.mylyn.bugzilla.ui (3.2.3.v20100217-0100-e3x) "Mylyn Bugzilla
Connector UI" [Active]
org.eclipse.mylyn.commons.core (3.2.3.v20100217-0100-e3x) "Mylyn Commons
Core" [Resolved] (3.2.0.v20090617-0100-e3x) "Mylyn Commons
Net" [Active]
org.eclipse.mylyn.commons.ui (3.2.3.v20100217-0100-e3x) "Mylyn Commons UI"
org.eclipse.mylyn.compatibility (3.2.3.v20100217-0100-e3x) "Java 5 Compatibility
Checker" [Resolved]
org.eclipse.mylyn.context.core (3.2.3.v20100217-0100-e3x) "Mylyn Context Core"
org.eclipse.mylyn.context.ui (3.2.3.v20100217-0100-e3x) "Mylyn Context UI"
org.eclipse.mylyn.discovery.core (3.2.3.v20100217-0100-e3x) "Mylyn Connector
Discovery Core" [Resolved]
org.eclipse.mylyn.discovery.ui (3.2.3.v20100217-0100-e3x) "Mylyn Connector
Discovery UI" [Resolved] (3.2.3.v20100217-0100-e3x) "Mylyn Help" [Starting]
org.eclipse.mylyn.ide.ant (3.2.3.v20100217-0100-e3x) "Mylyn Ant Bridge"
org.eclipse.mylyn.ide.ui (3.2.3.v20100217-0100-e3x) "Mylyn IDE UI" [Starting] (3.2.3.v20100217-0100-e3x) "Mylyn Java Tasks"
[Starting] (3.2.3.v20100217-0100-e3x) "Mylyn Java Bridge"
org.eclipse.mylyn.monitor.core (3.2.3.v20100217-0100-e3x) "Mylyn Monitor Core"
org.eclipse.mylyn.monitor.ui (3.2.3.v20100217-0100-e3x) "Mylyn Monitor UI"
org.eclipse.mylyn.resources.ui (3.2.3.v20100217-0100-e3x) "Mylyn Resources UI"
org.eclipse.mylyn.tasks.bugs (3.2.3.v20100217-0100-e3x) "Mylyn Bug Reporting"
org.eclipse.mylyn.tasks.core (3.2.3.v20100217-0100-e3x) "Mylyn Tasks Core"
org.eclipse.mylyn.tasks.ui (3.2.3.v20100217-0100-e3x) "Mylyn Tasks UI" [Active] (3.2.3.v20100217-0100-e3x) "Mylyn CVS Extensions"
[Resolved] (3.2.3.v20100217-0100-e3x) "Mylyn Team UI" [Active]
org.eclipse.mylyn.wikitext.confluence.core (1.1.3.v20100217-0100-e3x) "Mylyn
WikiText Confluence" [Resolved]
org.eclipse.mylyn.wikitext.confluence.ui (1.1.3.v20100217-0100-e3x) "Mylyn
WikiText Confluence UI" [Resolved]
org.eclipse.mylyn.wikitext.core (1.1.3.v20100217-0100-e3x) "Mylyn WikiText"
[Starting] (1.1.3.v20100217-0100-e3x) "Mylyn WikiText
Help UI" [Resolved]
org.eclipse.mylyn.wikitext.mediawiki.core (1.1.3.v20100217-0100-e3x) "Mylyn
WikiText MediaWiki" [Resolved]
org.eclipse.mylyn.wikitext.mediawiki.ui (1.1.3.v20100217-0100-e3x) "Mylyn
WikiText MediaWiki UI" [Resolved]
org.eclipse.mylyn.wikitext.tasks.ui (1.1.3.v20100217-0100-e3x) "Mylyn WikiText
Tasks UI" [Starting]
org.eclipse.mylyn.wikitext.textile.core (1.1.3.v20100217-0100-e3x) "Mylyn
WikiText Textile" [Resolved]
org.eclipse.mylyn.wikitext.textile.ui (1.1.3.v20100217-0100-e3x) "Mylyn WikiText
Textile UI" [Resolved]
org.eclipse.mylyn.wikitext.tracwiki.core (1.1.3.v20100217-0100-e3x) "Mylyn
WikiText TracWiki" [Resolved]
org.eclipse.mylyn.wikitext.tracwiki.ui (1.1.3.v20100217-0100-e3x) "Mylyn
WikiText TracWiki UI" [Resolved]
org.eclipse.mylyn.wikitext.twiki.core (1.1.3.v20100217-0100-e3x) "Mylyn
WikiText Twiki" [Resolved]
org.eclipse.mylyn.wikitext.twiki.ui (1.1.3.v20100217-0100-e3x) "Mylyn WikiText
Twiki UI" [Resolved]
org.eclipse.mylyn.wikitext.ui (1.1.3.v20100217-0100-e3x) "Mylyn WikiText UI"
org.eclipse.osgi (3.5.2.R35x_v20100126) "OSGi System Bundle" [Active] (3.2.0.v20090520-1800) "OSGi Release 4.2.0 Services"
org.eclipse.osgi.util (3.2.0.v20090520-1800) "OSGi Release 4.2.0 Utility Classes"
org.eclipse.platform (3.3.202.v201002111343) "Eclipse Platform" [Resolved]
org.eclipse.platform.doc.user (3.5.2.r352_v20091111-0800) "Eclipse Workbench
User Guide" [Resolved]
org.eclipse.rcp (3.5.0.v201002111343) "Eclipse RCP" [Starting] (3.5.1.r351_v20090708-0800) "Search Support" [Active]
org.eclipse.swt (3.5.2.v3557f) "Standard Widget Toolkit" [Resolved]
org.eclipse.swt.win32.win32.x86 (3.5.2.v3557f) "Standard Widget Toolkit for
Windows" [Resolved] (3.5.1.r35x_20100113-0800) "Team Support Core" [Active] (3.3.200.I20090430-0408) "CVS Team Provider Core"
[Active] (3.2.100.I20090508-2000) "CVS SSH Core" [Starting] (3.2.200.I20090508-2000) "CVS SSH2" [Starting] (3.3.202.r35x_20090930-0800) "CVS Team Provider UI"
[Active] (3.5.0.I20090430-0408) "Team Support UI" [Active]
org.eclipse.text (3.5.0.v20090513-2000) "Text" [Resolved]
org.eclipse.ui (3.5.2.M20100120-0800) "Eclipse UI" [Active]
org.eclipse.ui.browser (3.2.301.v20091215_35x) "Browser Support" [Starting]
org.eclipse.ui.cheatsheets (3.3.200.v20090526) "Cheat Sheets" [Active]
org.eclipse.ui.console (3.4.0.v20090513) "Console" [Active]
org.eclipse.ui.editors (3.5.0.v20090527-2000) "Default Text Editor" [Active]
org.eclipse.ui.externaltools (3.2.0.v20090504) "External Tools" [Starting]
org.eclipse.ui.forms (3.4.1.v20090714_35x) "Eclipse Forms" [Active]
org.eclipse.ui.ide (3.5.2.M20100113-0800) "Eclipse IDE UI" [Active]
org.eclipse.ui.ide.application (1.0.101.M20090826-0800) "Eclipse IDE UI
Application" [Resolved]
org.eclipse.ui.intro (3.3.2.v20100111_35x) "Welcome Framework" [Active]
org.eclipse.ui.intro.universal (3.2.300.v20090526) "Universal Welcome" [Active]
org.eclipse.ui.navigator (3.4.2.M20100120-0800) "Common Navigator View"
org.eclipse.ui.navigator.resources (3.4.1.M20090826-0800) "Navigator Workbench
Components" [Starting] (1.2.1.r35x_20090812-1200) "Internet Connection Management
UI" [Active]
org.eclipse.ui.presentations.r21 (3.2.100.M20091015-0930) "R21 Presentation Plugin" [Starting]
org.eclipse.ui.views (3.4.1.M20090826-0800) "Views" [Active]
org.eclipse.ui.views.log (1.0.100.v20090731) "Log View" [Starting] (3.5.0.I20090429-1800) "Tabbed Properties
View" [Starting]
org.eclipse.ui.win32 (3.2.100.v20090429-1800) "Eclipse UI Win32 Enhancements"
org.eclipse.ui.workbench (3.5.2.M20100113-0800) "Workbench" [Active]
org.eclipse.ui.workbench.compatibility (3.2.0.I20090429-1800) "Workbench
Compatibility" [Resolved]
org.eclipse.ui.workbench.texteditor (3.5.1.r352_v20100105) "Text Editor
Framework" [Active]
org.eclipse.update.configurator (3.3.0.v20090312) "Install/Update Configurator"
org.eclipse.update.core (3.2.300.v20090525) "Install/Update Core" [Active]
org.eclipse.update.core.win32 (3.2.100.v20080107) "Install/Update Core for
Windows" [Resolved]
org.eclipse.update.scheduler (3.2.200.v20081127) "Automatic Updates Scheduler"
org.eclipse.update.ui (3.2.201.R35x_v20090813) "Install/Update UI" [Starting]
org.eclipse.wst.common.core (1.1.201.v200806010600) "WST Common Core Plugin" [Starting]
org.eclipse.wst.common.emf (1.1.301.v200908181930) "EMF Utilities" [Starting]
org.eclipse.wst.common.emfworkbench.integration (1.1.301.v200908101600) "EMF
Workbench Edit Plug-in" [Starting]
org.eclipse.wst.common.environment (1.0.301.v200908101600) "Enviornment Plugin" [Starting]
org.eclipse.wst.common.frameworks (1.1.300.v200904160730) "Common
Frameworks" [Starting]
org.eclipse.wst.common.frameworks.ui (1.1.301.v200908101600) "WTP UI Plugin" [Starting]
org.eclipse.wst.common.infopop (1.0.100.v200805301550) "Common WST
infopops" [Starting]
org.eclipse.wst.common.modulecore (1.1.301.v201001252130) "Modulecore Plugin" [Starting]
org.eclipse.wst.common.project.facet.core (1.4.1.v200911141735) "Eclipse Faceted
Project Framework" [Starting]
org.eclipse.wst.common.snippets (1.1.301.v201001280015) "Snippets View"
org.eclipse.wst.common.ui (1.1.402.v200901262305) "Eclipse Base UI extensions"
org.eclipse.wst.common.uriresolver (1.1.301.v200805140415) "Common URI
Resolver Framework" [Starting]
org.eclipse.wst.dtd.core (1.1.300.v200904181727) "Structured Source DTD Core"
org.eclipse.wst.dtd.ui (1.0.400.v200904300717) "SSE DTD Source Editor"
org.eclipse.wst.dtd.ui.infopop (1.0.300.v200805140200) "DTD Editor infopops"
org.eclipse.wst.dtdeditor.doc.user (1.0.500.v200904292308) "DTD Editor
documentation" [Resolved]
org.eclipse.wst.internet.cache (1.0.301.v200805140020) "Cache URI Resolver Plugin" [Starting]
org.eclipse.wst.sse.core (1.1.402.v201001251516) "Structured Text Model"
org.eclipse.wst.sse.doc.user (1.1.0.v200805211530) "Structured text editor and
snippets documentation" [Resolved]
org.eclipse.wst.sse.ui (1.1.102.v200910200227) "Structured Source Editor"
org.eclipse.wst.sse.ui.infopop (1.0.200.v200805301545) "SSE infopops" [Resolved]
org.eclipse.wst.standard.schemas (1.0.200.v200902270215) "Standard Schemas and
DTDs" [Resolved]
org.eclipse.wst.validation (1.2.104.v200911120201) "Validation Framework"
org.eclipse.wst.validation.infopop (1.0.300.v200806041506) "WST validation
infopop plug-in" [Resolved]
org.eclipse.wst.validation.ui (1.2.104.v200904232035) "Validation Framework UI"
org.eclipse.wst.xml.core (1.1.402.v201001222130) "Structured Source XML
Model" [Starting]
org.eclipse.wst.xml.ui (1.1.2.v201001222130) "Eclipse XML Editors and Tools"
org.eclipse.wst.xml.ui.infopop (1.0.300.v200805140200) "XML infopops"
org.eclipse.wst.xmleditor.doc.user (1.0.600.v200901231300) "XML editor"
org.eclipse.wst.xsd.core (1.1.401.v200903020335) "XSD Core Plugin" [Starting]
org.eclipse.wst.xsd.ui (1.2.204.v200909021537) "XML Schema Editor" [Starting]
org.eclipse.wst.xsdeditor.doc.user (1.0.700.v200905182240) "XML schema editor"
org.eclipse.xsd (2.5.0.v200906151043) "XSD Model" [Starting]
org.eclipse.xsd.edit (2.5.0.v200906151043) "XSD Edit" [Starting]
org.hamcrest.core (1.1.0.v20090501071000) "Hamcrest Core Library of Matchers"
org.junit (3.8.2.v20090203-1005) "JUnit Testing Framework" [Resolved]
org.junit4 (4.5.0.v20090824) "JUnit Testing Framework Version 4" [Resolved]
org.mortbay.jetty.server (6.1.15.v200905151201) "Jetty Server" [Resolved]
org.mortbay.jetty.util (6.1.15.v200905182336) "Jetty Utilities" [Resolved]
org.sat4j.core (2.1.1.v20090825) "SAT4J Core" [Resolved]
org.sat4j.pb (2.1.1.v20090825) "SAT4J Pseudo" [Resolved]
*** User Preferences:
#Fri Mar 26 09:41:50 EST 2010
load.eclipse.org_technology_epp_packages_galileo/nickname=EPP Packages
ipse/name=Bundle pool
ml version\="1.0" encoding\="UTF-8" standalone\="no"?>\r\n<vmSettings
id\="1269556301121" javadocURL\="http\://"
name\="jre6" path\="C\:\\Program
wnload.eclipse.org_technology_epp_packages_galileo/nickname=EPP Packages
w=<?xml version\="1.0" encoding\="UTF-8"?>\r\n<VariablesViewMemento
=metadata listener dropins
wnload.eclipse.org_eclipse_updates_3.5/nickname=The Eclipse Project Updates
_JADE projects
org.eclipse.debug.ui.DebugView=<?xml version\="1.0" encoding\="UTF8"?>\r\n<DebugViewMemento
load.eclipse.org_eclipse_updates_3.5/nickname=The Eclipse Project Updates
load.eclipse.org_tools_mylyn_update_e3.4/nickname=Mylyn for Eclipse 3.4 and 3.5
New|10.0|0|WINDOWS|1|0|0|0|0|0|0|0|0|1|0|0|0|0|Courier New;
version\="1.0" encoding\="UTF-8"
rtifact listener dropins
ipse_p2_org.eclipse.equinox.p2.core_cache/name=download cache
wnload.eclipse.org_tools_mylyn_update_e3.4/nickname=Mylyn for Eclipse 3.4 and
*** Current Install Configuration:
Install configuration:
Last changed on 26/12/2010
Location: C:\eclipse
Profile timestamp: 1269556260080
Installable Units in the profile:
Id: a.jre.javase, Version: 1.6.0
Id:, Version: 4.0.1.v20090822
Id: com.jcraft.jsch, Version: 0.1.41.v200903070017
Id: config.a.jre.javase, Version: 1.6.0
Id:, Version:
Id: it.fbk.sra.ejade, Version: 0.8.0
Id: javax.servlet, Version: 2.5.0.v200806031605
Id: javax.servlet.jsp, Version: 2.0.0.v200806031607
Id: javax.xml, Version: 1.3.4.v200902170245
Id: org.apache.ant, Version: 1.7.1.v20090120-1145
Id: org.apache.commons.codec, Version: 1.3.0.v20080530-1600
Id: org.apache.commons.codec, Version: 1.3.0.v20100106-1700
Id: org.apache.commons.el, Version: 1.0.0.v200806031608
Id: org.apache.commons.httpclient, Version: 3.1.0.v20080605-1935
Id: org.apache.commons.lang, Version: 2.3.0.v200803061910
Id: org.apache.commons.logging, Version: 1.0.4.v200904062259
Id: org.apache.jasper, Version: 5.5.17.v200903231320
Id: org.apache.lucene, Version: 1.9.1.v20080530-1600
Id: org.apache.lucene.analysis, Version: 1.9.1.v20080530-1600
Id: org.apache.xerces, Version: 2.9.0.v200909240008
Id: org.apache.xml.resolver, Version: 1.2.0.v200902170519
Id: org.apache.xml.serializer, Version: 2.7.1.v200902170519
Id: org.eclipse.ant.core, Version: 3.2.101.v20091110_r352
Id: org.eclipse.ant.ui, Version: 3.4.2.v20091204_r352
Id:, Version: 3.5.2.r35x_20100113-0800
Id:, Version: 3.5.0.I20090430-0408
Id:, Version: 1.0.0.I20090430-0408
Id: org.eclipse.core.boot, Version: 3.1.100.v20080218
Id: org.eclipse.core.commands, Version: 3.5.0.I20090525-2000
Id: org.eclipse.core.contenttype, Version: 3.4.1.R35x_v20090826-0451
Id: org.eclipse.core.databinding, Version: 1.2.0.M20090819-0800
Id: org.eclipse.core.databinding.beans, Version: 1.2.0.I20090525-2000
Id: org.eclipse.core.databinding.observable, Version: 1.2.0.M20090902-0800
Id:, Version: 1.2.0.M20090819-0800
Id: org.eclipse.core.expressions, Version: 3.4.101.R35x_v20100209
Id: org.eclipse.core.filebuffers, Version: 3.5.0.v20090526-2000
Id: org.eclipse.core.filesystem, Version: 1.2.1.R35x_v20091203-1235
Id: org.eclipse.core.filesystem.win32.x86, Version: 1.1.0.v20080604-1400
Id:, Version: 3.4.100.v20090429-1800
Id:, Version: 1.2.1.r35x_20090812-1200
Id:, Version: 1.0.0.I20080909
Id: org.eclipse.core.resources, Version: 3.5.2.R35x_v20091203-1235
Id: org.eclipse.core.resources.compatibility, Version: 3.4.1.R35x_v20100113-0530
Id: org.eclipse.core.resources.win32.x86, Version: 3.5.0.v20081020
Id: org.eclipse.core.runtime, Version: 3.5.0.v20090525
Id: org.eclipse.core.runtime.compatibility, Version: 3.2.0.v20090413
Bundles in the system:
Id:, Version: 4.0.1.v20090822, Location:
Id: com.jcraft.jsch, Version: 0.1.41.v200903070017, Location:
Id: it.fbk.sra.ejade, Version: 0.8.0, Location:
Id: javax.servlet, Version: 2.5.0.v200806031605, Location:
Id: javax.servlet.jsp, Version: 2.0.0.v200806031607, Location:
Id: javax.xml, Version: 1.3.4.v200902170245, Location:
*** Security Configuration:
Providers (9):
Provider: SUN, Version: 1.6, Class:
Description: SUN (DSA key/parameter generation; DSA signing; SHA-1, MD5
digests; SecureRandom; X.509 certificates; JKS keystore; PKIX CertPathValidator;
PKIX CertPathBuilder; LDAP, Collection CertStores, JavaPolicy Policy;
JavaLoginConfig Configuration)
Services (23):
Service: MessageDigest, Algorithm: SHA-512, Class:$SHA512
Service: Configuration, Algorithm: JavaLoginConfig, Class:
Service: Signature, Algorithm: NONEwithDSA, Class:$RawDSA
Aliases: RawDSA
Service: CertPathBuilder, Algorithm: PKIX, Class:
ImplementedIn: Software
ValidationAlgorithm: RFC3280
Service: CertStore, Algorithm: LDAP, Class:
ImplementedIn: Software
LDAPSchema: RFC2587
Service: MessageDigest, Algorithm: SHA, Class:
Aliases: SHA1
ImplementedIn: Software
Service: CertificateFactory, Algorithm: X.509, Class:
Aliases: X509
ImplementedIn: Software
Service: MessageDigest, Algorithm: SHA-384, Class:$SHA384
Service: KeyPairGenerator, Algorithm: DSA, Class:
Aliases: OID.1.2.840.10040.4.1
ImplementedIn: Software
KeySize: 1024
Service: KeyStore, Algorithm: JKS, Class:$JKS
ImplementedIn: Software
Service: KeyStore, Algorithm: CaseExactJKS, Class:$CaseExactJKS
Service: MessageDigest, Algorithm: SHA-256, Class:
Service: CertStore, Algorithm:, Class:
ImplementedIn: Software
Service: KeyFactory, Algorithm: DSA, Class:
Aliases: OID.1.2.840.10040.4.1
ImplementedIn: Software
Service: SecureRandom, Algorithm: SHA1PRNG, Class:
ImplementedIn: Software
Service: CertPathValidator, Algorithm: PKIX, Class:
ImplementedIn: Software
ValidationAlgorithm: RFC3280
Service: CertStore, Algorithm: Collection, Class:
ImplementedIn: Software
Service: MessageDigest, Algorithm: MD5, Class:
ImplementedIn: Software
Service: AlgorithmParameters, Algorithm: DSA, Class:
Aliases: OID.1.2.840.10040.4.1
ImplementedIn: Software
Service: Policy, Algorithm: JavaPolicy, Class:
Service: MessageDigest, Algorithm: MD2, Class:
Service: Signature, Algorithm: SHA1withDSA, Class:$SHA1withDSA
Aliases: DSAWithSHA1
ImplementedIn: Software
KeySize: 1024
Service: AlgorithmParameterGenerator, Algorithm: DSA, Class:
Aliases: OID.1.2.840.10040.4.1
ImplementedIn: Software
KeySize: 1024
Provider: SunRsaSign, Version: 1.5, Class:
Description: Sun RSA signature provider
Services (8):
Service: Signature, Algorithm: SHA512withRSA, Class:$SHA512withRSA
Aliases: OID.1.2.840.113549.1.1.13
Service: Signature, Algorithm: MD5withRSA, Class:$MD5withRSA
Aliases: 1.2.840.113549.1.1.4
Service: KeyFactory, Algorithm: RSA, Class:
Aliases: 1.2.840.113549.1.1
Service: Signature, Algorithm: SHA1withRSA, Class:$SHA1withRSA
Aliases: OID.1.2.840.113549.1.1.5
Service: Signature, Algorithm: SHA384withRSA, Class:$SHA384withRSA
Aliases: OID.1.2.840.113549.1.1.12
Service: Signature, Algorithm: MD2withRSA, Class:$MD2withRSA
Aliases: 1.2.840.113549.1.1.2
Service: KeyPairGenerator, Algorithm: RSA, Class:
Aliases: 1.2.840.113549.1.1
Service: Signature, Algorithm: SHA256withRSA, Class:$SHA256withRSA
Aliases: OID.1.2.840.113549.1.1.11
Provider: SunJSSE, Version: 1.6, Class:
Description: Sun JSSE provider(PKCS12, SunX509 key/trust factories, SSLv3,
Provider: XMLDSig, Version: 1.0, Class:
Description: XMLDSig (DOM XMLSignatureFactory; DOM KeyInfoFactory)
Services (11):
Service: KeyInfoFactory, Algorithm: DOM, Class:
Service: TransformService, Algorithm:, Class:
Aliases: XPATH
MechanismType: DOM
Service: TransformService, Algorithm:, Class:
MechanismType: DOM
Service: TransformService, Algorithm:, Class:
Aliases: BASE64
MechanismType: DOM
Service: TransformService, Algorithm:, Class:
MechanismType: DOM
Service: TransformService, Algorithm:, Class:
MechanismType: DOM
Service: XMLSignatureFactory, Algorithm: DOM, Class:
Service: TransformService, Algorithm:, Class:
Aliases: XPATH2
MechanismType: DOM
Service: TransformService, Algorithm:, Class:
MechanismType: DOM
Service: TransformService, Algorithm:, Class:
Aliases: XSLT
MechanismType: DOM
Service: TransformService, Algorithm:, Class:
MechanismType: DOM
Provider: SunPCSC, Version: 1.6, Class:
Description: Sun PC/SC provider
Services (1):
Service: TerminalFactory, Algorithm: PC/SC, Class:$Factory
Provider: SunMSCAPI, Version: 1.6, Class:
Description: Sun's Microsoft Crypto API provider
Services (12):
Service: Signature, Algorithm: SHA512withRSA, Class:$SHA512
Service: Signature, Algorithm: MD5withRSA, Class:$MD5
Service: Signature, Algorithm: SHA1withRSA, Class:$SHA1
Service: Cipher, Algorithm: RSA, Class:
SupportedPaddings: PKCS1PADDING
SupportedModes: ECB
Service: SecureRandom, Algorithm: Windows-PRNG, Class:
Service: KeyStore, Algorithm: Windows-ROOT, Class:$ROOT
Service: Signature, Algorithm: SHA384withRSA, Class:$SHA384
Service: KeyPairGenerator, Algorithm: RSA, Class:
KeySize: 1024
Service: Signature, Algorithm: SHA256withRSA, Class:$SHA256
Service: Cipher, Algorithm: RSA/ECB/PKCS1Padding, Class:
Service: Signature, Algorithm: MD2withRSA, Class:$MD2
Service: KeyStore, Algorithm: Windows-MY, Class:$MY
Appendix F: Agent ALMA and Cogito Ontology
Code Java Code
ALMA Agent
// Set up an objective Alma agent
public class Alma extends Agent{
protected void setup(){
//Notify its presence in experiment
System.out.println(" Internal" +getAID().getName() +" is
present and waiting ");
// Register the Alma_Legal ontology for this agent
private Codec almac = new SLCodec();
private Ontology AlmaLegal = ALMA_LegalOntology.getInstance();
{ getContentManager() .registerLanguage(almac);
getContentManager() .registerOntology(AlmaLegal);}
private boolean finished =false;
//Alma Agent Communication behaviour
//waiting for objective agent interaction to be internalised
private class intern extends CyclicBehaviour{
public void action() {
// TODO Auto-generated method stub
ACLMessage obj = myAgent.receive();
if (obj !=null){
//Objective message received Process it
String title = obj.getContent();
// flag internalisation
OBJ Agent
import jade.core.behaviours.CyclicBehaviour;
import jade.core.behaviours.*;
import jade.lang.acl.ACLMessage;
// Set up an objective Alma agent
public class Obj extends Agent{
protected void setup(){
//Notify its presence in experiment
System.out.println("Objective Agent" +getAID().getName()
+" is present and waiting ");
// Register the Alma_Legal ontology for this agent
private Codec almac = new SLCodec();
private Ontology AlmaLegal = ALMA_LegalOntology.getInstance();
{ getContentManager() .registerLanguage(almac);
getContentManager() .registerOntology(AlmaLegal);}
private boolean finished =false;
//Objective Agent Communication behaviour
//waiting for external interaction to be internalised to Alma
private class intern extends CyclicBehaviour{
public void action() {
// TODO Auto-generated method stub
ACLMessage obj = myAgent.receive();
if (obj !=null){
//Objective message received Process it
String title = obj.getContent();
// flag internalisation
SUBJ Agent
import jade.core.behaviours.*;
import jade.lang.acl.ACLMessage;
// Set up a subjective Alma agent
public class Sub extends Agent{
protected void setup(){
//Notify its presence in experiment
System.out.println("Subjective Agent"
+getAID().getName() +" is present and waiting ");
// Register the Alma_Legal ontology for this agent
private Codec almac = new SLCodec();
private Ontology AlmaLegal = ALMA_LegalOntology.getInstance();
{ getContentManager() .registerLanguage(almac);
getContentManager() .registerOntology(AlmaLegal);}
private boolean finished =false;
//Subjective Agent Communication behaviour
//waiting for internal interaction with Alma to be
private class intern extends CyclicBehaviour{
public void action() {
// TODO Auto-generated method stub
ACLMessage obj = myAgent.receive();
if (obj !=null){
//Objective message received Process it
String title = obj.getContent();
// flag internalisation
Cogito Ontology
/** file:
* @author ontology bean generator
* @version 2010/07/1, 10:27:23
public class CogitoOntology extends jade.content.onto.Ontology
public static final String ONTOLOGY_NAME = "Cogito";
// The singleton instance of this ontology
private static ReflectiveIntrospector introspect = new
private static Ontology theInstance = new CogitoOntology();
public static Ontology getInstance() {
return theInstance;
public static final String KNOW="Know";
public static final String INTERNALISE="Internalise";
public static final String EXTERNALISE="Externalise";
public static final String ALMA_COGITO="ALMA_Cogito";
public static final String SUBJECTIVE="Subjective";
public static final String OBJECTIVE="Objective";
public static final String
public static final String
public static final String
public static final String
* Constructor
private CogitoOntology(){
super(ONTOLOGY_NAME, BasicOntology.getInstance());
try {
// adding Concept(s)
// adding AgentAction(s)
AgentActionSchema cogito_Build_ConceptSchema = new
AgentActionSchema cogito_Change_SelfSchema = new
add(cogito_Change_SelfSchema, Cogito.Cogito_Change_Self.class);
AgentActionSchema cogito_InternaliseSchema = new
add(cogito_InternaliseSchema, Cogito.Cogito_Internalise.class);
AgentActionSchema cogito_ExternaliseSchema = new
add(cogito_ExternaliseSchema, Cogito.Cogito_Externalise.class);
// adding AID(s)
ConceptSchema objectiveSchema = new ConceptSchema(OBJECTIVE);
add(objectiveSchema, Cogito.Objective.class);
ConceptSchema subjectiveSchema = new ConceptSchema(SUBJECTIVE);
add(subjectiveSchema, Cogito.Subjective.class);
ConceptSchema almA_CogitoSchema = new
add(almA_CogitoSchema, Cogito.ALMA_Cogito.class);
// adding Predicate(s)
PredicateSchema externaliseSchema = new
add(externaliseSchema, Cogito.Externalise.class);
PredicateSchema internaliseSchema = new
add(internaliseSchema, Cogito.Internalise.class);
PredicateSchema knowSchema = new PredicateSchema(KNOW);
add(knowSchema, Cogito.Know.class);
// adding fields
// adding name mappings
// adding inheritance
}catch (java.lang.Exception e) {e.printStackTrace();}
Appendix G: PiersonvPost Legal Case
Pierson v. Post Case copy
3 Cai. R. 175; 1805 N.Y. LEXIS 311
August, 1805, Decided
[**1] THIS was an action of trespass on the case commenced in a justice's court, by the
present defendant against the now plaintiff.
The declaration stated that Post, being in possession of certain dogs and hounds under
his command, did, "upon a certain wild and uninhabited, unpossessed and waste land,
called the beach, find and start one of those noxious beasts called a fox," and whilst there
hunting, chasing and pursuing the same with his dogs and hounds, and when in view
thereof, Pierson, well knowing the fox was so hunted and pursued, did, in the sight of
Post, to prevent his catching the same, kill and carry it off. A verdict having been
rendered for the plaintiff below, the defendant there sued out a certiorari, and now
assigned for error, that the declaration and the matters therein contained were not
sufficient in law to maintain an action.
Judgment of reversal.
Animals Ferae Naturae-- What Gives Right of Property in Trespass.
Pursuit alone gives no right of property in animals feroe naturoe, therefore an action will
not lie against a man for killing and taking one pursued by, and in the view of, the person
who originally found, started, chased it, and was on the point of seizing it.
Occupancy in wild animals can be acquired only by possession, but such possession does
not signify manucaption, though it must be of such a kind as by nets, snares or other
means, as to so circumvent the creature that he cannot escape. Citations--Just. Inst., lib. 2,
tit. 1, sec. 13; Fleta, lib. 3, ch. 2, p. 175; Bracton, lib. 2, ch. 1, p. 8; Puffendorf, lib. 4, ch. 6,
sec. 2, 10; Grotius, lib. 2, ch. 8, sec. 3, p. 309; 11 Mod., 74-130; 3 Salk., 9.
COUNSEL: Mr. Sanford, for the now plaintiff. It is firmly settled that animals, feroe
naturoe, belong not to anyone. If, then, Post had not acquired any property in the fox,
when it was killed by Pierson, he had no right in it which could be the subject of injury.
As, however, a property may be gained in such an animal, [**2] it will be necessary to
advert to the facts set forth, to see whether they are such as could give a legal interest in
the creature, that was the cause of the suit below. Finding, hunting, and pursuit, are all
that the plaint enumerates. To create a title to an animal feroe naturor, occupancy is
indispensable. It is the only mode recognized by our system. 2 Black. Com. 403.
The reason of the thing shows it to be so. For whatever is not appropriated by positive
institutions, can be exclusively possessed by natural law alone. Occupancy is the sole
method this code acknowledges. Authorities are not wanting to this effect. Just. lib. 2, tit.
1, sec. 12. "Feroe igitur bestioe, simulatque ab aliiquo captoe fuerint jure gentium statim
illius esse incipiunt." There must be a taking; and even that is not in all cases sufficient,
for in the same section he observes, "Quicquid autem corum ceperis, eo usque tuum esse
intelligitur, donec tua custodia coercetur; cum vero tuam evaserit custodiam, et in
libertatem naturalem sese receperit, tuam esse desinit, et rursus occumpantis fit." It is
added also that this natural liberty may be regained even if in sight of the pursuer, "ita sit,
ut difficilis [**3] sit ejus persecutio." In section 13, it is laid down, that even wounding
will not give a right of property in an animal that is unreclaimed. For, not withstanding
the wound, "multa accidere soleant ut eam non capias," and "non aliter tuam esse quam si
eam ceperis." Fleta (b. 3, p. 175) and Bracton (b. 2, ch. 1, p. 86) are inunison with the
Roman law-giver. It is manifest, then, from the record, that there was no title in Post, and
the action, therefore, not maintainable.
Mr. Colden, contra. I admit, with Fleta, that pursuit alone does not give a right of
property in animals feroe naturoe, and I admit also that occupancy is to give a title to
them. But, then, what kind of occupancy? And here I shall contend it is not such as is
derived from manucaption alone. In Puffendorf's Law of Nature and of Nations (b. 4, ch.
4, sec. 5, n. 6, by Barbeyrac), notice is taken of this principle of taking possession. It is
there combatted, nay, disproved; and in b. 4, ch. 6, sec. 2, n. 2. lbid. sec. 7, n. 2,
demonstrated that manucaption is only one of many means to declare the intention of
exclusively appropriating that which was before in a state of nature. Any continued act
which does this, [**4] is equivalent to occupancy. Pursuit, therefore, by a person who
starts a wild animal, gives an exclusive right whils it is followed. It is all the possession the
nature of the subject admits; it declares the intention of acquiring dominion, and is as
much to be respected as manucaption itself. The contrary idea, requiring actual taking,
proceeds, as Mr. Barbeyrac observes, in Puffendorf (b. 4, ch. 6, sec. 10), on a "false
notion of possession."
Mr. Sanford, in reply. The only authority relied on is that of an annotator. On the
question now before the court, we have taken our principles from the civil code, and
nothing as been urged to impeach those quoted from the author referred to.
[*177] TOMPKINS, J., delivered the opinion of the court:
This cause comes before us on a return to a certiorari directed to one of the justices of
Queens County.
The question submitted by the counsel in this cause for our determination is, whether
Lodowick Post, by the pursuit with his hounds in the manner alleged in his declaration,
acquired such a right to, or property in, the fox as will sustain an action against Pierson
[**5] for killing and taking him away?
The cause was argued with much ability by the counsel on both sides, and presents for
our decision a novel and nice question. It is admitted that a fox is an animal fertoe
naturoe, and that property in such animals is acquired by occupancy only. These
admissions narrow the discussion to the simple question of what acts amount to
occupancy, applied to acquiring right to wild animals.
If we have recourse to the ancient writers upon general principles of law, the judgment
below is obviously erroneous. Justinian's Institutes (lib. 2, tit. 1, sec. 13), and Fleta (lib. 3,
ch. 2, p. 175), adopt the principle, that pursuit alone vests no property or right in the
huntsman; and that even pursuit, accompanied with wounding, is equally ineffectual for
that purpose, unless the animal be actually taken. The same principle is recognized by
Breton (lib. 2, ch. 1, p. 8).
Puffendorf (lib. 4, ch. 6, sec. 2 and 10) defines occupancy of beasts feroe naturoe, to be
the actual corporeal possession of them, and Bynkershock is cited as coinciding in this
definition. It is indeed with hesitation that Puffendor affirms that a wild beast mortally
wounded or greatly [**6] maimed, cannot be fairly intercepted by another, whilst the
pursuit of [*178] the person inflicting the wound continues. The foregoing authorities are
decisive to show that mere pursuit gave Post no legal right to the fox, but that he became
the property of Pierson, who intercepted and killed him.
It, therefore, only remains to inquire whether there are any contrary principles or
authorities, to be found in other books, which ought to induce a different decision. Most
of the cases which have occurred in England, relating to property in wild animals, have
either been discussed and decided upon the principles of their positive statute regulations,
or have arisen between the huntsman and the owner of the land upon which beasts feroe
naturoe have been apprehended; the former claiming them by title of occupancy, and the
latte ratione soli. Little satisfactory aid can, therefore, be derived from the English
Barbeyrac, in his notes on Puffendorf, does not accede to the definition of occupancy
by the latter, but, on the contrary, affirms that actual bodily seizure is not, in all cases,
necessary to constitute possession of wild animals. He does not, however, describe [**7]
the acts which, according to his ideas, will amount to an appropriation of such animals to
private use, so as to exclude the claims of all other persons, by title of occupancy, to the
same animals; and he is far from averring that pursuit alone is sufficient for that purpose.
To a certain extent, and as far as Barbeyrac appears to me to go, his objections to
Puffendorf's definition of occupancy are reasonable and correct.
That is to say, that actual bodily seizure is not indispensable to acquire right to, or
possession of, wild beasts; but that, on the contrary, the mortal wounding of such beasts,
by one not abandoning his pursuit, may, with the utmost propriety, be deemed
possession of him; since thereby the pursuer manifests an unequivocal intention of
appropriating the animal to his individual use, has deprived him of his natural liberty, and
brought him within his certain control.
So, also, encompassing and securing such animals with nets and toils, or otherwise
intercepting them in such a manner as to deprive them of their natural liberty, and render
escape impossible, may justly be deemed to give possession of them to those persons
who, by their industry and labor, have used [**8] such means of apprehending them.
Barbeyrac seems to have adopted and had in view in his notes, [*179] the more accurate
opinion of Grotius, with respect to occupancy. That celebrated author (lib. 2, ch. 8, sec. 3,
p. 309), speaking of occupancy, proceeds thus: "Requiritur autem corporalis quoedam
possessio ad dominium adipiscendum; atque ideo, vulnerasse non sufficit." But in the
following section he explains and qualifies this definition of occupancy: "Sed possessio
illa potest non solis manibus, sed instrumentis, ut decipulis, ratibus, laqueis dum duo
adsint; primum ut ipsa instrumenta sint in nostra potestate, deinde ut fera, ita inclusa sit,
ut exire inde nequeat." This qualification embraces the full extent of Barbeyrac's objection
to Puffendorf's definition, and allows as great a latitude to acquiring property by
occupancy, as can reasonably be inferred from the words or ideas expressed by Barbeyrac
in his notes. The case now under consideration is one of mere pursuit, and presents no
circumstances or acts which can bring it within the definition of occupancy by
Puffendorf, or Grotius, or the ideas of Barbeyrac upon that subject.
The case cited from 11 Mod. 74, 130, [**9] I think clearly distinguishable from the
present; inasmuch as there the action was for maliciously hindering and disturbing the
plaintiff in the exercise and enjoyment of a private franchise; and in the report of the
same case (3 Salk. 9), Holt, Ch. J., states, that the ducks were in the plaintiff's decoy pond,
and so in his possession, from which it is obvious the court laid much stress in their
opinion upon the plaintiff's possession of the ducks, ratione soli.
We are the more readily inclined to confine possession or occupancy of beasts feroe
naturoe, within the limits prescribed by the learned authors above cited, for the sake of
certainty, and preserving peace and order in society. If the first seeing, starting or
pursuing such animals, without having so wounded, circumvented or ensnared them, so
as to deprive them of their natural liberty, and subject them to the control of their
pursuer, should afford the basis of actions against others for intercepting and killing
them, it would prove a fertile source of quarrels and litigation.
However uncourteous or unkind the conduct of Pierson towards Post, in this instance,
may have been, yet this act was productive of no injury [**10] or damage for which a
legal remedy [*180] can be applied. We are of opinion the judgment below was
erroneous, and ought to be reversed.
LIVINGSTON, J. My opinion differs from that of the court. Of six exceptions, taken
to the proceedings below, all are abandoned except the third, which reduces the
controversy to a single question.
Whether a person who, with his own hounds, starts and hunts a fox on waste and
uninhabited ground, and is on the point of seizing his prey, acquires such an interest in
the animal as to have a right of action against another, who in view of the huntsman and
his dogs in full pursuit, and with knowledge of the chase, shall kill and carry him away.
This is a knotty point, and should have been submitted to the arbitration of sportsmen,
without poring over Justinian, Fleta, Bracton, Puffendorf, Locke, Barbeyrac, or
Blackstone, all of whom have been cited: they would have had no difficulty in coming to
a prompt and correct conclusion. In a court thus constituted, the skin and carcass of poor
Reynard would have been properly disposed of, and a precedent set, interfering with no
usage or custom which the experience of ages has sanctioned, and which must be [**11]
so well known to every votary of Diana. But the parties have referred the question to our
judgment, and we must dispose of it as well as we can, from the partial lights we possess,
leaving to a higher tribunal the correction of any mistake which we may be so
unfortunate as to make. By the pleadings it is admitted that a fox is a "wild and noxious
beast." Both parties have regarded him, as the law of nations does a pirate, "hostem
humani generis," and although "de mortuis nil nisi bonum" be a maxim of our
profession, the memory of the deceased has not been spared. His depredations on
farmers and on barnyards, have not been forgotten; and to put him to death wherever
found, is allowed to be meritorious, and of public benefit. Hence it follows, that our
decision should have in view the greatest possible encouragement to the destruction of an
animal, so cunningand ruthless in his career. But who would keep a pack of hounds; or
what gentleman, at the sound of the horn, and at peep of day, would mount his steed,
and for [*181] hours together, "sub jove frigido," or a vertical sun, pursue the windings of
this wily quadruped, if, just as night came on, and his stratagems [**12] and strength
were nearly exhausted, a saucy intruder, who had not shared in the honors or labors of
the chase, were permitted to come in at the death, and bear away in triumph the object of
pursuit? Whatever Justinian may have thought of the matter, it must be recollected that
his code was compiled many hundred years ago, and it would be very hard indeed, at the
distance of so many centuries, not to have a right to establish a rule for ourselves. In his
day, we read of no order of men who made it a business, in the language of the
declaration in this cause, "with hounds and dogs to find, start, pursue, hunt, and chase,"
these animals, and that, too, without any other motive than the preservation of Roman
poultry; if this diversion had been then in fashion, the lawyers who composed his
institutes, would have taken care not to pass it by, without suitable encouragement. If
anything, therefore, in the digests or pandects shall appear to militate against the
defendant in error, who, on this occasion, was the fox hunter, we have only to say
tempora mutantur; and if men themselves change with the times, why should not laws
also undergo an alteration?
It may be expected, however, by [**13] the learned counsel, that more particular notice
be taken of their authorities. I have examined them all, and feel great difficulty in
determining, whether to acquire dominion over a thing, before in common, it be
sufficient that we barely see it, or know where it is, or wish for it, or make a declaration of
our will respecting it; or whether, in the case of wild beasts, setting a trap, or lying in wait,
or starting, or pursuing, be enough; or if an actual wounding, or killing, or bodily tact and
occupation be necessary. Writers on general law, who have favored us with their
speculations on these points, differ on them all; but, great as is the diversity of sentiment
among them, some conclusion must be adopted on the question immediately before us.
After mature deliberation, I embrace that of Barbeyrac as the most rational and least
liable to objection. If at liberty, we might imitate the courtesy of a certain emperor, who,
to avoid giving [*182] offense to the advocates of any of these different doctrines,
adopted a middle course, and by ingenious distinctions, rendered it difficult to say (as
often happens after a fierce and angry contest) to whom the palm of victory belonged.
[**14] He ordained, that if a beast be followed with large dogs and hounds, he shall
belong to the hunter, not to the chance occupant; and in like manner, if he be killed or
wounded with a lance or sword; but if chased with beagles only, then he passed to the
captor, not to the first pursuer. If slain with a dart, a sling, or a bow, he fell to the hunter,
if still in chase, and not to him who might afterwards find and seize him.
Now, as we are without any municipal regulations of our own, and the pursuit here, for
aught that appears on the case, being with dogs and hounds of imperial stature, we are at
liberty to adopt one of the provisions just cited, which comports also with the learned
conclusion of Barbeyrac, that property in animals feroe naturoe may be acquired without
bodily touch or manucaption, provided the pursuer be within reach, or have a reasonable
prospect (which certainly existed here) of taking what he has thus discovered an intention
of converting to his own use.
When we reflect also that the interest of our husbandmen, the most useful of men in any
community, will be advanced by the destruction of a beast so pernicious and incorrigible,
we cannot greatly err in saying [**15] that a pursuit like the present, through waste and
unoccupied lands, and which must inevitably and speedily have terminated in corporeal
possession, or bodily seisin, confers such a right to the object of it, as to make any one
awrong-doer who shall interfere and shoulder the spoil. The justice's judgment ought,
therefore, in my opinion, to be affirmed.
Judgment of reversal. n1
Wild bees in a bee-tree belong to the owner of the soil where the tree stands. Ferguson
v. Miller, 1 Cow. 243. Though another discover the bees, and obtain license from the
owner to take them, and mark the tree with the initials of his own name, this does not
confer the ownership upon him, until he has taken actual possession of the bees. ld.
If he omit to take such possession, the owner of the soil may give the same license to
another, who may take the bees without being liable to the first finder. Id. The two
parties, both having license, the one who takes possession first, acquires the title. Id. Bees
are animals feroe naturoe, but when hived and reclaimed, a qualified property may be
acquired in them. Gillett v. Mason, 7 Johns.
16. If a person find a tree, containing a hive of bees, on the land of another, and mark
the tree, he does not thereby reclaim the bees, and vest a right of property in himself; and
cannot maintain an action for carrying away the bees and honey. Id.
Though property in animal feroe naturoe may be acquired by occupancy, or by
wounding it, so as to bring it within the power or control of the pursuer; yet, if after
wounding the animal and continuing the pursuit of it until evening, the hunter abandons
the pursuit, though his dogs continue chase, he acquires no property in the animal. Buster
v. Newkirk, 20 Johns. 75; N. Y. Dig., Vol. I., p.
106, et seq.
End of Footnotes
Agent Oriented Software Engineering
Agent Communication Language
Artificial Intelligence
Agent Language Mediated Activity
JADE Agent Management System
Activity Theory
Agent Unified Modelling Language
Connected Limited Device Configuration
Distributed Artificial Intelligence
Defence Advanced Research Project Agency
JADE Directory Facilitator
Foundation for Intelligent Physical Agents
Graphic User Interface
Human Agent Interface
Human Computer Interface
Hypertext Transmission Protocol
Internet Inter-ORB Protocol
Java Agent Development Framework
Java Development Kit
Java Virtual Memory
Knowledge Query Markup Language
Knowledge Software Engineering
Lightweight Extensible Authentication Protocol
Lesser GNU Public Language
Legal Knowledge Interchange Format
Mobile Information Device Profile
Multiple Agents Systems
Object Request Broker
Ontology Web Language
JADE Remote Monitoring Agent
Remote Procedure control
Extensible Markup Language