* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download Knowledge Representation in Competence Management using First
Survey
Document related concepts
Intelligence explosion wikipedia , lookup
Soar (cognitive architecture) wikipedia , lookup
Human–computer interaction wikipedia , lookup
Logic programming wikipedia , lookup
Ethics of artificial intelligence wikipedia , lookup
Ecological interface design wikipedia , lookup
Machine learning wikipedia , lookup
Computer Go wikipedia , lookup
Existential risk from artificial general intelligence wikipedia , lookup
Philosophy of artificial intelligence wikipedia , lookup
Personal knowledge base wikipedia , lookup
Transcript
Knowledge Representation in Competence Management using First Order Predicate Logic for Machine Learning: A Mathematical Approach to Artificial Intelligence Application Prof Pooja Tripathi Associate Professor ,IPEC, Research Scholar BIT MESRA Ranchi. To whom all the contacts may be sent: [email protected], [email protected] Dr Jayanthi Ranjan Professor, Institute of Management Technology , Ghaziabad. Dr Tarun Pandeya Dr S.L. Gupta Professor, BIT Mesra, Ranchi India Abstract - Almost every day we come across with some situation that needs attention where we apply our general problem solving behavior to address it. In most of the cases we apply our intelligence and sometimes the integrated effort by a group of experts to address the issue. Competence Assessment and performance management of the organization is a complex decision making process. To identify a right candidate for the job various factors are to be considered in the form personality, ability, skills and knowledge and to be assessed in the candidate. Competence Management and Performance Assessment have emerged as a multidisciplinary complex science involving several of skills related to number of technology and the behavioral indicators. The paper explores the method for the information gathering from competence experts and choosing the proper formats by the knowledge engineers to represent the information. The Knowledge is represented in form of data bases, facts and rules using suitable data base management system, programming languages as AI tools to develop the production rules like if and then. First Order Predicate Logic is an advanced form of propositional calculus, which is the simplest form of knowledge representation that comprises of statements. The paper discusses the FOPL representation for the Competence Management System. Keywords - First Order Predicate Logic, Knowledge Representation, Machine Learning. I. INTRODUCTION Almost every day we come across with some situation that needs attention where we apply our general problem solving behavior to address it. In most of the cases we apply our intelligence and sometimes the integrated effort by a group of experts to address the issue. For a fixed situation, we normally arrive at a consensus that can be converted to an algorithm that a machine can apply and solve the problem. This is like your income tax calculation, some statistical inferences or making some predictions based on some mathematical models. Observing the The Eighth International Conference on eLearning for Knowledge-Based Society, 23-24 February 2012, Thailand 1 Prof Pooja Tripathi, Dr Jayanthi Ranjan, Dr Tarun Pandeya, and Dr S.L. Gupta capabilities of a modern computer a new area of application has emerged in which a machine adopts the problem solving behaviour of a human. This area is known as Artificial Intelligence that takes new approaches to solve problem to a very large domain. Artificial Intelligence, otherwise known as AI, is the study and development of intelligent machines capable of performing complex tasks that require thought and behavior normally associated with human intelligence. Computer programs are a common area of specialization in this branch of science. Artificial Intelligence adapts characteristics of human problemsolving skills and then applies them as algorithms easily comprehended by computer systems. Such systems are routinely and widely used today in robotics, corporations, militaries and homes around the world. One application of this technology is knowledge based systems where knowledge of human experts is utilized by the machines to take decisions. The power of machine comes from the power of its knowledge base. Information acquired from experts is stored in it in different forms and formats chosen by the knowledge engineers. Knowledge is represented in form of data bases, facts and rules using suitable data base management system, programming languages or AI tools using production rules like if and then. Some of the benefits of IFTHEN rules are that they are modular, each defining a relatively small and, at least in principle, independent piece of knowledge. New rules may be added and old ones deleted usually independently of other rules. A Thinking Machine = A Knowledge Base + An Inference Engine These rules that are used for knowledge representation is used for machine learning. Machines are trained with these rules with different form of knowledge representation. Machine learning programs examine the database to automatically determine the rules (Michalski, R. S.,1983). Machine learning was first used in agriculture to develop rules for diagnosing soybean diseases (PLANT/ds) (Michalski, R. S. 1980.). A microcomputerbased, machine-learning system has been developed for agricultural problems (Fermanian, T. W.,1988). This system was first used to generate rules for a grass identification system (Weeder). First Order Predicate Logic could also be utilized as a tool for machine learning for decision making in human resource and development. 1.1 Theory about First Order Predicate Logic First Order Predicate Logic is an advanced form of propositional calculus, which is the simplest form of knowledge representation that comprises of statements. For example All men are Mortal and Socrates is Mortal infers Socrates is mortal. First-order logic permits reasoning about the propositional connectives (as in propositional logic) and also about quantification ("all" or "some"). First-order logic is symbolized reasoning in which each sentence, or statement, is broken down into a subject and a predicate. The predicate modifies or defines the properties of the subject. A sentence in first-order logic is written in the form Px or P(x), where P is the predicate and x is the subject, represented as a variable. Complete sentences are logically combined and manipulated according to the same rules as those used in Boolean algebra. First order logic is very well understood, and has a sound mathematical foundation. It combines an expressive language with a sound method of inference in that language. Therefore FOL has been immensely popular in AI's attempt to construct intelligent programs. First-order predicate logic is composed of statements that are assumed to be true. The statements are composed of: atoms (symbols), predicates (a function with one or more atomic arguments), two substatements joined by a conjunction, disjunction, or implication, a negated substatement, and Special Issue of the International Journal of the Computer, the Internet and Management, Vol. 19 No. SP2, February, 2012 2 Knowledge Representation in Competence Management using First Order Predicate Logic for Machine Learning: A Mathematical Approach to Artificial Intelligence Application a statement with an existential or universal quantifier (in this case, atoms in the statements can be replaced by variables in the quantifier). The predicate calculus includes a wider range of entities. It permits the description of relations and the use of variables. It also requires an understanding of quantification. The language of predicate calculus requires Variables and Constants ---these include the logical constants Predicate ---these relate a number of entities. This number is usually greater than one. A predicate with one argument is often used to express a property e.g. sun(hot) may represent the statement that ``the sun has the property of being hot''. Sentence ---a formula with no free variables is a sentence. Two informal examples to illustrate quantification follow: Utility of First Order Predicate Logic (FOPL) First-order logic can be useful in the creation of computer programs. It is also of interest to researchers in artificial intelligence (AI). There are more powerful forms of logic, but first-order logic is adequate for most everyday reasoning. Most systems use rules (e.g., IF a THEN b) to represent domain knowledge; however, more complicated representations do exist. Another issue of knowledge representation is uncertainty. Some knowledge bases will allow rules to be expressed with a degree of uncertainty. The inference engine will then combine these uncertainties to give the final solution a confidence level. A third feature of the knowledge base is the ability to express concepts in symbolic terms (e.g., IF the stove is HOT THEN set dial to MEDIUM) instead of in numeric terms. This feature allows more abstract concepts to be expressed. The inference engine is responsible for finding the proper knowledge in the knowledge base to develop a solution. For rule-based systems, the inference engine combines (i.e., chains) rules together using rules of logic. 1.2 Objectives of the Research Acquisition of Knowledge from Domain Expert on Competence mapping and assessment Development of Production Rules for Competence Identification Representation of these rules using First Order Predicate Logic Utilization of these rules for machine learning II. RESEARCH METHODOLOGY To identify a right candidate for the job the main factors to be considered are whether the competencies in the form personality, ability, skills and knowledge are found in candidate or not. If the competencies are found in the candidate, then system identifies the candidate with the help of production rules developed for the identification of the right candidate for performing the right job or if candidate is not identified, then it identifies the competencies gap on the basis of production rules developed for the given job in terms of the competencies requirements. A complete detail of the candidate identification process is acquired from the published reports and experts. Production rules are developed using the acquired information and by the help of experts. First Order Predicate Logic is utilized for the representation of these rules. First Order Predicate Logic is also utilized for developing the inference mechanism. The knowledge so represented by the First Order Predicate Logic is utilized for machine learning. The inference mechanism so developed is also applied for machine learning. A suitable programming language is applied for the process development. 2.1 Knowledge Acquiring through Questionnaire Technique The questionnaire used, began with a brief introduction about the research study The Eighth International Conference on eLearning for Knowledge-Based Society, 23-24 February 2012, Thailand 3 Prof Pooja Tripathi, Dr Jayanthi Ranjan, Dr Tarun Pandeya, and Dr S.L. Gupta which specified that the researcher’s interest in their perceptions of what they think the competent faculty should have. The survey was designed to make it as easy, convenient, less time-consuming and as interesting as possible. The data was collected through the personal meet, focus group interviews. The questionnaire consisted of two parts: the first section gathers some simple demographic data like age, education, gender, teaching experience and other work responsibilities and so on, followed by the second section which consists of a list of competence attributes to be evaluated by the participant. This section entails attributes of the job itself as well as the environment and the physical location of the work place. The final questionnaire was consisting of 58 items which were chosen from the original pool of 73 items.A 5-point scale ranging from 1: ‘Least Important’ to 5: ‘Most important’ was used to study participants’ assessments of individual attributes and values. On an average the survey took about 12-15 minutes to complete. 2.2 Knowledge Representation in Competence Management and its importance Competence Management and Performance Assessment have emerged as a multidisciplinary complex science involving several of skills related to number of technology and the behavioral indicators. Every discipline has its own physiology and needs a specific environment and technology for its proper growth. Human Resource research involves a large number of discipline and areas that generates competence profiling, competence collection, competence assessment, and evaluation techniques for the same. If the knowledge management will be done by using IT based tools, it will be permanent information that could be accessible for the employee and the management whosoever is interested in the area. It will minimize duplication, better research planning and management by better access and expansion. III. FUNCTIONAL DESCRIPTION OF DEVELOPED SYSTEM The system has been developed for the competence management and performance assessment of the faculty members and can be used by the educational institutions. The development of the proposed system includes the development of different subsystems. The above section describes the different sub-systems in the developed system as well as its functionality. 1. The knowledge base: Developed to enable the expert or knowledge engineer to enter and save domain ontology and domain rules in an efficient and easy way. 2. Web page interface of the proposed system: For using purposes includes entering ontology, knowledge, and the use of the expert system by entering the options and get the output results with reasoning too. 3. The entire control and operation of the system is done by the inference engine; that is developed using ASP dot net (ASP.NET)and the MS SQL server; which handles the knowledge in format of XML to get the result from the XML file (Knowledge base) that stores the knowledge rules. The main roles of the inference engine are summarized as: It applies the expert domain knowledge to what is known about the present situation to determine new information about the domain. The inference engine is the mechanism that connects the user inputs in the form of answers to the questions to the rules of knowledge base and further continues the session to come to conclusions. This process leads to the solution of the problem. The inference engine also identifies the rules of the knowledge base used to get decision from the system and also forms the decision tree. The user submits the query through the interface to the system. The system generates a series of questionnaire and takes the input from the user. The inference engine accepts Special Issue of the International Journal of the Computer, the Internet and Management, Vol. 19 No. SP2, February, 2012 4 Knowledge Representation in Competence Management using First Order Predicate Logic for Machine Learning: A Mathematical Approach to Artificial Intelligence Application these queries and takes the input from the user and uses this dynamic information together with the existing knowledge on competence management and the performance assessment. The contents of the knowledge base are used to derive the conclusion. The inference mechanism is carried out in four phases. Select --→ Identify --→ Match -→Execute. During the match stage the contents of working memory is compared with the facts and rules stored in the knowledge base. When consistent matches are found the rules are again compared on the basis of the input given by the user and finally one of the rule is executed on selection of the image. IV. CONCLUSION As result of using the developed system, a cultural shift occurred within the educational institution. Faculty members became more accountable to teacher candidate competency attainment and teacher candidates became more attuned to becoming an effective prospective teacher. The Competence and Performance expert system implemented quality principles, utilized the value added approach to management and systematically used data that yielded positive results; it has become a model for other teacher education programs. REFERENCES [1] Michalski, R. S., J. G. Carbonell, and T. M. Mitchell. 1983.Machine learning: an artificial intelligence approach. MorganKaufmann Publishers, Inc., Los Altos, CA. [2] Michalski, R. S. 1980. Learning by being told and learning from examples: an experimental comparison of the two methods of knowledge acquisition in the context of developing an expert system for soybean disease diagnosis. International Journal of Policy Analysis and Information Systems 4:125-161. [3] Fermanian, T. W., R. S. Michalski, and B. Katz. 1988. AgAssistant - A tool for developing expert systems. Page 252-271. Knowledge Based Systems in Agriculture – Prospects for Application. edited by Deutsche LandwirtschaftsGesellschaft, Frankfurt, Germany. [4] Liepins, G., R. Goeltz, and R. Rush. 1990. Machine learning techniques for natural resource data analysis. AI Applications in Natural Resource Management 4(3):9-18. [5] Dill, D. E. 1990. An electronic model for determining value and predicting sale price of Holstein cattle sold at public auction. Ph.D. Thesis, University of Illinois, Urbana. [6] Saarenmaa, H., N. D. Stone, L. J. Folse, J. M. Packard, W. E. Grant, M. E. Makela, and R. N. Coulson. 1988. An artificial intelligence modelling approach to simulating animal/habitat interactions. Ecological Modelling 44:125-141. [7] Beck, H. W., P. Jones, and J. W. Jones. 1989. SOYBUG: an expert system for soybean insect pest management. Agricultural Systems 30:269286. [8] Roach, J. W., R. S. Virkar, M. J. Weaver, and C. R. Drake. 1985. POMME: a computer-based consultation system for apple orchard management using PROLOG. Expert Systems 2(2):56-69. Fig.1 Overall Competence based Management The Eighth International Conference on eLearning for Knowledge-Based Society, 23-24 February 2012, Thailand 5