Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Semantic Web Applications © Copyright 2010 Dieter Fensel and Kathaina Siorpaes www.sti-innsbruck.at 1 Where are we? # Title 1 Introduction 2 Semantic Web Architecture 3 Resource Description Framework (RDF) 4 Web of data 5 Generating Semantic Annotations 6 Storage and Querying 7 Web Ontology Language (OWL) 8 Rule Interchange Format (RIF) 9 Reasoning on the Web 10 Ontologies 11 Social Semantic Web 12 Semantic Web Services 13 Tools 14 Applications www.sti-innsbruck.at 2 Agenda 1. Motivation 2. Technical solutions and illustrations 1. 2. 3. 4. 5. 6. Application for Semantic Indexing and Semantic Portals (Watson) Application for description, discovery and selection (Search Monkey) ACTIVE case study: British Telecom INSEMTIVES case studies LARKC case study SOA4All case study 3. Extensions 4. Summary 5. References www.sti-innsbruck.at 3 3 MOTIVATION www.sti-innsbruck.at 4 4 Motivation • A wide variety of applications of semantic technologies. • Novel technology is often validated in real world case studies. • Example: – Company X wants to improve their knowledge management system by semantic technology. – Company Y produces virtual worlds and wants to annotate multimedia elements in these games. – Etc. • Common scenarios: – – – – – Data integration Knowledge management Indexing Annotation and enrichment Discovery (search) www.sti-innsbruck.at 5 5 TECHNICAL SOLUTION AND ILLUSTRATIONS www.sti-innsbruck.at 6 6 Application for Semantic Indexing and Semantic Portals: Dr. Watson http://watson.kmi.open.ac.uk www.sti-innsbruck.at 7 7 Applications for Semantic Indexing and Semantic Portals • Web already offers topic-specifigc portals and generic structured directories like Yahoo! or DMOZ • With semantic technologies such portals could: – use deeper categorization and use ontologies – integrate indexed sources from many locations and communities – provide different structured views on the underlying information • Example application: Watson www.sti-innsbruck.at 8 8 Watson – What is it? • Watson is a gateway for the semantic web • Provides efficient access point to the online ontologies and semantic data • Is developed at the Knoledge Media Institute of the Open Universit in Milton Keynes, UK *) *) Source: http://watson.kmi.open.ac.uk/Overview.html www.sti-innsbruck.at 9 9 Watson – How does it work? • Watson collects available semantic content on the Web • Analyzes it to exstract useful metadata and indexes it • Implements efficient query facilities to acess the data *) *) Source: http://watson.kmi.open.ac.uk/Overview.html www.sti-innsbruck.at 10 10 Watson – Features at a Glance • Attempt to provide high quality semantic data by ranking available data • Efficient exploration of implicit and explicit relations between ontologies • Selecting only relevant ontology modules by extraciting it from the whole ontology • Different interfaces for querying and navigation as well as different levels of formalization www.sti-innsbruck.at 11 11 Watson – An example Search for movie and director www.sti-innsbruck.at Resulting ontologies 12 12 SearchMonkey http://developer.yahoo.com/search monkey/ www.sti-innsbruck.at 13 13 Applications for description, discovery and selection • • • Category of applications the are closely related to semantic indexing and knowledge management Applications mainly for helping users to locate a resource, product or service meeting their needs Example application: SearchMonkey www.sti-innsbruck.at 14 14 SearchMonkey – What is it? • Search monkey is a framework for creating small applications that enhance Yahoo! Search results • Additional data, structure, images and links may be added to search results • Yahoo provides meta-data *) *) Source: http://developer.yahoo.com/searchmonkey/smguide/index.html www.sti-innsbruck.at 15 15 SearchMonkey – An example application • IMDB Infobar • Enhance searches for imdb.com/name and imdb.com/title • Adds information about the searched movie and links to the search result • May be added individually to enhance once search results www.sti-innsbruck.at 16 16 SearchMonkey – How does it work? • • • • Applications use two types of data services: custom ones and ones provided by Yahoo! Yahoo! Data services include: – Indexed Web Data – Indexed Semantic Web Data – Cached 3rd party data feeds Custom data services provide additional, individual data SearchMonkey application processes the provided data and presents it *) *) Source http://developer.yahoo.com/searchmonkey/smguide/data.html www.sti-innsbruck.at 17 17 SearchMonkey – Ontologies used • • Common vocabularies used: Friend of a Friend( foaf), Dublin Core (dc), VCard(vcard), VCalendar(vcal), etc. SearchMonkey specific: – searchmonkey-action.owl: for performing actions as e.g. comparing prices of items – searchmonkey- commerce.owl: for displaying various information collected about businesses – searchmonkey-feed.owl: for displaying information from a feed – searchmonkey-job.owl: for displaying information found in job descriptions or recruitment postings – searchmonkey-media.owl: for displaying information about different media types – searchmonkey-product.owl: for displaying information about products or manufacturers – searchmonkey-resume.owl: for displaying information from a CV • SearchMonkey does not support reasoning of OWL data www.sti-innsbruck.at 18 18 ACTIVE case study www.active-project.eu Slides by Ian Thurlow, BT www.sti-innsbruck.at 19 19 Background • Background for ACTIVE – 80% of corporate information is unstructured – task switching is a productivity killer – 80% of business people use 2 or more devices and 20% use 4 or more – most knowledge worker activity is not based on formal processes • In addition, knowledge workers tend to be: – – – – overloaded with information (from multiple sources) interact with multiple systems geographically dispersed under pressure to reduce costs (respond and deliver better, quicker) www.sti-innsbruck.at 21 Overall aims of the case study • To improve the effectiveness of BT’s knowledge workers through the use (and evaluation) of ACTIVE technology • Give people the information/knowledge they want (filter out what they don’t need) – based on people’s context and their priorities – re-use existing information/knowledge • Put people in touch with other people (relevant to their current work) – make knowledge-sharing easy and natural www.sti-innsbruck.at 22 Overall aims of the case study (cont.) • Guide people through their everyday tasks – identify informal processes, e.g. when creating a bid response, training somebody new to the area – learning from previous experiences – maximise re-use of solutions • Reduce task switching • Provide a useful and robust ‘knowledge workspace’ www.sti-innsbruck.at 23 The knowledge workspace Context: • • • • • • Interrupts: • • • • • • • • • • • • • • E-mail (multiple accounts) Voice mail Schedule Task list IM SMS News items Stock market Weather Security alarm Bank alerts Travel news Media … www.sti-innsbruck.at Interest profile Device type Connectivity Time, date, location Current tasks Community… ACTIVE Technology Features • Filtering information • Learning your interests • Learning your knowledge processes • Modelling your context • Learning your priorities Knowledge Workspace: • Prioritisation of interrupts • Automated support for knowledge processes • Concise, timely, relevant information • Context and device sensitive 24 Research challenges • How do we learn and exploit user context to give users access to information (and knowledge): – – – – that they want when they want it in the form in which they want it whilst mitigating information overload! • How do we share information more effectively – to support other people undertaking similar tasks – without interrupting people unnecessarily – without overloading people with information www.sti-innsbruck.at 25 Research challenges (cont.) • How do we learn and exploit informal knowledge processes both to guide users through those processes and to streamline them – suggest actions to users based on previous ways of carrying out a process – simplifying processes, making suggestions to users accordingly • How do we measure the benefits of Active (technology)? – e.g. efficiency of users, ease of access to information, user satisfaction? www.sti-innsbruck.at 26 Deployment challenges • Engage the participants – BT Retail sales workforce • Sales specialists, technical consultants, sales consultants • Keep participants interested (for 3 years) • Target: 200+ people (M36) • BT sales workforce/sales specialists - very busy people – access to their time will be limited – they will not tolerate anything which hinders their work • applications and tools will need to be useful and robust – people already interact with multiple systems (be careful of introducing others) • can not just use the BT case study as a ‘test bed’ for all active technologies • selective use of ACTIVE technology (to meet business needs) • Integration of Active technology www.sti-innsbruck.at 27 BT Systems (a sample) • Core tools – MS Internet Explorer, Excel, Outlook, PowerPoint, Project, Visio, Word • Communications – Instant messaging, mobile phone, SMS, SoftPhone (VoIP), BT MeetMe (conference), MS LiveMeeting, desk phone • Information sources – BT Corporate Viewer (customer information), BT directory, Intellact (corporate information/news), Sales Zone (product information), Offer Factory (Corporate Proposal & bid support documents) • Process – LiveLink (document storage), One View (customer order and information tracking – Siebel based), Salesforce.com (pilot system evaluation for tracking customer deals) www.sti-innsbruck.at 28 INSEMTIVES case studies www.insemtives.eu www.sti-innsbruck.at 29 29 Idea Realizing the Semantic Web by encouraging millions of end-users to create semantic content. 30 www.sti-innsbruck.at 5/9/2017 www.insemtives.eu 30 What • Bridging the gap between human and computational intelligence in semantic content authoring. • Methodologies, methods, tools for the large-scale creation of semantics – Driven by ideas from incentive theory and participatory design. – Optimally combine human input and automatic techniques. • Wide range of content types (text, multimedia, Web services). • Case studies addressing the most important issues of semantic content creation projects. www.sti-innsbruck.at www.insemtives.eu 31 Why • More and more information is available on the Web. The information overflow is unmanageable. • Semantic technologies help to make sense of this huge amount of information. • BUT: Many tasks related to semantic content creation are human-driven and can not be carried out automatically. • Limited involvement of users in the Semantic Web. • Incentive structures are not in place for semantic content authoring. www.sti-innsbruck.at www.insemtives.eu 32 How www.sti-innsbruck.at www.insemtives.eu 33 Expected outcomes • A unified methodology for authoring semantic data. • Incentive mechanisms for semantic content creation. • Design guidelines for tools. www.sti-innsbruck.at www.insemtives.eu 34 Expected outcomes (cont) • Models and methods for the creation of lightweight, structured knowledge. – Bootstraping through the extraction of contextual knowledge. – Converge of semantics. – Linking semantic content. – Semantic search. www.sti-innsbruck.at www.insemtives.eu 35 Expected outcomes (cont) • A semantic content management platform for the storage and retrieval of user-generated content, including methods for supporting the lifecycle of this content. www.sti-innsbruck.at www.insemtives.eu 36 Expected outcomes (cont) • A toolkit implementing guidelines and incentive mechanisms for ontology development and annotation of different types of media. – Generic games toolkit and games. – Semi-automatic annotation tools. – Bootstrapping tools. – Search and navigation tools. www.sti-innsbruck.at www.insemtives.eu 37 Expected outcomes (cont) • 3 case studies for evaluation of INSEMTIVES technology in realworld settings. – Different types of communities of users. – Different types of information items. – Different types of semantic content. www.sti-innsbruck.at www.insemtives.eu 38 OKEnterprise • • • In corporate environments, important information is often lost. Okenterprise is a social network for corporate knowledge management in Telefonica. We will apply INSEMTIVES technology to this network in order to generate and share new knowledge among coworkers. www.sti-innsbruck.at www.insemtives.eu 39 Virtual worlds • Media producers and companies face the lack of reliable metadata for the huge collections of assets they produce. • In this case study, we will apply incentive methods to the virtual world “Tiny planets” to semi-automate the creation of descriptive metadata. www.sti-innsbruck.at www.insemtives.eu 40 Web service annotation • The lack of rich descriptions beyond their current syntactical interface hampers the automatic retrieval of Web services on the Internet. • The case study will apply INSEMTIVES technology to facilitate user-provided annotation of Web services. www.sti-innsbruck.at www.insemtives.eu 41 Potential impact • Massive amounts of useful semantic content which can enable the uptake of semantic technology through the development of application producing real added value for the Semantic Web and for industrial adopters. – Production of digital resources easier and more cost-effective – Enhanced search of digital resources • Case studies solving real world problems – PGP: multimedia annotation – Seekda: annotation of Web services – Telefonica: semantically enhanced corporate knowledge management 42 www.sti-innsbruck.at 5/9/2017 www.insemtives.eu 42 LARKC case study: Urban computing (www.larkc.eu) Slides by LARKC project wiki www.sti-innsbruck.at 43 43 Today Cities’ Challenges • Our cities face many challenges • How can we redevelop existing neighbourhoods and business districts to improve the quality of life? • How can we create more choices in housing, accommodating diverse lifestyles and all income levels? • How can we reduce traffic congestion yet stay connected? • How can we include citizens in planning their communities rather than limiting input to only those affected by the next project? • How can we fund schools, bridges, roads, and clean water while meeting short-term costs of increased security? www.sti-innsbruck.at 44 Urban Computing as a Way to Address those challenges www.sti-innsbruck.at 45 45 The reasoning challenge Coping with zillions of facts •Heterogeneous •Inconsistent •Unbounded •Coming in rapid, continuous, time-varying (burst) streams •Correlated but un-related Real-time requirements • All data cannot be taken into consideration at the same time • Need for abstracting rough data in meaningful facts • Need for selecting the relevant ones • Need for parallel inference and query processing Real-time requirements • Graceful approximation of results while applying selection and abstraction techniques www.sti-innsbruck.at 46 Short Term CEFRIEL’s Traffic Predictor • • CEFRIEL together with Milano Municipality has develop a Traffic Predictor (TP) for emergency vehicle routing in the Milano fair area The objective of TP (2 years long for some 60 PM effort) was to simulate real traffic in a metropolitan area in order to achieve: – Short-term (i.e.:10-15 min) traffic conditions on the whole area – Emergency Vehicle guidance support system – Long-term (i.e.: 6-48 hours) traffic conditions on the whole area Network and Traffic Data Models / heuristics Microscopic Simulator Traffic Flow HighLights and Control plan Macroscopic Simulator Decision Simulation results compare Data Control Center Network model www.sti-innsbruck.at 47 47 Input data and simulation • • • Input data: – static • A detailed (1 meter resolution) vectorial map of the 15,3 Km2 of the Milano fair area • All vertical and horizontal traffic signs • Traffic lights and their daily and weekly timing • Parking lots and major destinations • Distribution of driving styles among drivers – Dynamic • 75 traffic detectors in the Milano fair area that generate a stream of data updated every 5 minutes – Historical • 3 months of data are kept for statistical purposes Simulation – Micro-simulation of position an speed for a maximum of 40.000 “standard” vehicles – Macro-simulation of number of vehicles and average speed per segment Output data: – Number of vehicles and average speed for each segment (junction-to-junction) in the next 10-15 minutes (meaningful up to 48 hours) www.sti-innsbruck.at 48 Micro-simulation Macro-simulation 48 Micro-scopic simulation www.sti-innsbruck.at 49 49 The problem for LarKC • • Once the cars are in the area the two simulators handles them But – How many car will enter the area? – Which are their destinations? • TP uses the historic data and simple heuristics for each traffic detector – People exist later today # – t Something is blocking the traffic, people will use different street # – • t And a couple of others Can LarKC do better? www.sti-innsbruck.at 50 50 Comparing and Contrastic LarKC and CEFRIEL Traffic Predictor • CEFRIEL fine tuned TP by hand untill it was able to “reasonably” predict both short-term and long-term traffic conditions in the area of Milano fair • However predictions are not always good due to many factors – A traffic detector may have been put on a road that officially has only 1 lane, but people normally use the lane as it was a two lane – A traffic detector may have been put on a road that officially has to 2 lanes, but people park in one of the lane – Traffic lights timing can be wrong – And many others • The TP project collected 3 month of historical data, CEFRIEL could negotiate with the stake holders to share those data with LarKC. • Proposal: We take 2 months of data as input and then challenge LarKC to perform better than TP hand tuned simulator other the last month of data www.sti-innsbruck.at 51 51 Short Term Saltlux’s Ubiquitous City Service Period Project 2007. 03 ~ 2007.06 (4months) Intelligent Car Navigation Service Work Traffic control application for intelligent car navigation Ontology modeling for u-city services Development for reasoning technology to cover city-scale Development of service scenarios for u-city Business Modeling • Scope •Business process analysis www.sti-innsbruck.at KB Modeling • Ontology • Reasoning Rule Analysis • Reasoning Engine • Architecture • Related systems Pilot System • Infra & reasoning S/W installation • Applications • POC verification 52 Background: U-City Project in Korea Songdo Organization: New Songdo City Development LLC(NSC) Area: Songdo(International Songo Business Compound) 5,619,834 m2 Period: 2003 ~ 2014 Cost : 1 billion euro www.sti-innsbruck.at • Korea is a leader in building social spaces online and they connect back to the real world very well • Ubiquitous technologies will let us strengthen this linkage by: - merging online social networks with offline social - linking online and offline events and information • Asia Trade Tower(2006 ~ 2010. 12) • Convention Center & Hotel(2006 ~ 2008) • Apartments & Stores(2006 ~ 2014) • Central Park(~ 2008.11) • Ecotarium(2007. 2 ~ 2009. 12) • Waterfront Park • International Hospital • Golf Course (2007. 4 ~ 2009. 4) 53 Objective & Scope: Traffic Control System U-City is an integrated, intelligent and innovative new city-making service that works through city domain convergence based on ubiquitous computing and information communication technology. It includes system integration, operation and all services except devices. www.sti-innsbruck.at 54 Use case Scenario: Intelligent Navigation 1. Normal Path 2. Detour by Accident at the starting point 3. Detour by Accident on a road www.sti-innsbruck.at 55 Architecture & Ontology Modeling Identified key concept through domain competency questions and used a traffic agent with U-city ontology and rules Type Building Web application RoadAction Interface Agent Traffic Agent Reasoner 3 CompleteEquipmentCo mpany 1 SOR (with OntoBroker 4.3) 1 Hospital 1 InsuranceCompany 1 LevelOfService 6 Rule 8 PoliceStation 1 RecommendationBasis 1 TrafficAccidentAgencySt at TrafficAccidentStat *Level of Service www.sti-innsbruck.at 228 PlannedEventStat Road Ontology 88 FireStation Link Reasoning Core Knowledge Base Creator 20 CarSituation Coordination LOS* Total TrafficEventTime 30 4 432 2 56 SOA4All case study www.soa4all.eu Marc Richardson, John Davies BT www.sti-innsbruck.at 57 57 Overview • BT's acquisition of Ribbit and its implications • Telco role(s) in the open services world • Scenario – Storyboard Overview – Actors involved – Sequence of activities in composition example www.sti-innsbruck.at 58 Objectives • Ribbit – Services toolkit for accessing and using BTs exposed ‘capabilities’ (VOIP, SMS, etc.), – allowing 3rd party developers to create mash-ups with other services • Case study creating future Ribbit infrastructure based on SOA4All technology – Semantic technology – improved service discovery and composition – Web 2.0 – building a community of developers – Context-aware support for service providers and consumers • Key objective – improve the process of creating novel Ribbit-based services – reduce cost and time of using and combining the services www.sti-innsbruck.at 59 BT Ribbit Acquisition • The BT use case was originally based on the Web21c web-based SDK, released in July 2007 • BT acquired Ribbit in October 2008 for $100m • Ribbit is a platform for building web-based Telco applications, offering similar (but more mature) services than the Web21c SDK • The Ribbit community is well established and has more developers than Web21c did (>10000 vs >1000) www.sti-innsbruck.at 60 Ribbit Advantages • Shows major commitment from BT in pursuing Telco services over the web – Accelerate transformation – Access to scarce talent • Ribbit is implementing a full set of lightweight RESTful services – Web21c SDK had more complicated WSDL with WS-Security – Ribbit better aligned with goals of SOA4All – Opportunity to use the MicroWSMO format developed in SOA4All project • Good links established with key people in Ribbit – Strong interest in SOA4All project and our SoftTelco work – Organised internal SoftTelco symposium attracting great interest www.sti-innsbruck.at 61 Telco role(s) in the Service World Open service web platform and tools 62 www.sti-innsbruck.at 62 1-Sided Business Model (Traditional Telco) € € Suppliers www.sti-innsbruck.at Telco Customers 63 2-Sided Business Model € Suppliers € Newspapers Readers € Advertisers www.sti-innsbruck.at 64 2-Sided/Multisided Business Model Service Providers • • € REVENUE SHARE PAY FOR PUBLICATION … Tools Community Security Billing Telco (Retail) Content Government Retailers Developers Service Wrap € Service Consumers Telco € • • • PAYG SUBSCRIPTION PREMIUM Advertising www.sti-innsbruck.at 65 From Telco to SoftTelco • Significant shift in business model – – – – N-sided business models Multiple revenue profiles (PAYG, subscription, …) Disruption to current pricing structures Price sensitivity determines the price balance between the market sides – Goal is to maximise revenue overall – Example: newspapers typically have lower cover prices to maximise readership, which maximises advertising revenues www.sti-innsbruck.at 66 SoftTelco – The Long Tail Short Tail [Traditional] Demand Long Tail [SoftTelco] Ribbit focus is on large number of niche applications in ‘the tail’ Traditional focus on mainstream products & markets New growth opportunities Number of Products www.sti-innsbruck.at 67 67 Scenarios Lightweight Service Creators • Easy to use interface for creating simple Telco apps using Ribbit Services • Some semi-automatic composition • Semantic service discovery • Web 2.0 community for encouraging innovation sharing BT Service Resellers • Reselling BT (and other) white label services • More complex compositions of BT services, internal company services, and OSS • Service management requirements (QOS, SLAs, fault handling) • Telco Domain ontologies (e.g. SID, eTom) www.sti-innsbruck.at 68 BT Lightweight Scenario Example: Meet Friends Composite Service • A service that allows you to organise a meeting with a group of friends at short notice – Get list of friends from social networking site (e.g. Facebook) – Find out which ones are in the area using Ribbit location service – Find out weather and travel information for proposed meeting venue from 3rd party – Send out invite and directions using Ribbit SMS www.sti-innsbruck.at 69 BT Meet Friends Mash-up CreateEvent - location - number of people - invite message - location preference ProvideContacts [Mock-up Service] LocationOfContact ListLocal SendMessage Bad weather, directions to the alternative venue Good weather, directions to the venue weather? Filter people SendMessage TravelRoute Weather 70 www.sti-innsbruck.at 70 Business Reseller Scenario • Example: Group SMS Company • A Company which allows you to ring a number and leave a message. – Converted to text and sent out as SMS messages to a to groups of people who are subscribed to that service – Company buys bulk rate SMS service from Ribbit – A free service to subscribers, subsidised with own advertising model – User profile and location are used to contextualise SMS advert – e.g. Traffic Alert SMS service, or Weather Warning Service • Much faster, more agile from concept to market 71 www.sti-innsbruck.at 71 Business Reseller Scenario Group SMS Company Authenticate Get User Profiles Group voice message Mobile Number Get Contextual Ads Create Messages Get Location Speech to Text Messages with contextual Ads Send Messages Update User Account Reseller’s services BT services 72 www.sti-innsbruck.at 72 EXTENSIONS www.sti-innsbruck.at 73 73 Extensions • • More information about tools and applications of semantic technologies is available at http://semanticweb.org/wiki/Tools Further EU projects with case studies: http://www.stiinnsbruck.at/research/projects/ www.sti-innsbruck.at 74 74 SUMMARY www.sti-innsbruck.at 75 75 Summary • Common application domains for semantic technology: – – – – – • • • • • • Data integration Knowledge management Indexing Annotation and enrichment Discovery (search) Dr. Watson semantic content crawler Yahoo! SearchMonkey Semantic technology for knowledge workers in ACTIVE Incentives for semantic technology Reasoning for urban computing in LARKC SOA in British Telecom www.sti-innsbruck.at 76 76 References • Mandatory reading – Project websites on use cases – http://www.w3.org/2001/sw/Europe/reports/chosen_demos_ratio nale_report/hp-applications-selection.html – http://www.readwriteweb.com/archives/10_semantic_apps_to_w atch_one_year_later.php www.sti-innsbruck.at 77 77 References • Further reading – – – – – – – – – – http://www.w3.org/2001/sw/Europe/reports/chosen_demos_rationale_report/hp-applications-selection.html http://dbpedia.org/About http://watson.kmi.open.ac.uk/Overview.html http://semanticweb.org/wiki/Main_Page http://simile.mit.edu/wiki/Piggy_Bank http://swaml.berlios.de/ http://developer.berlios.de/projects/swaml/ http://rdfs.org/sioc/spec/ http://watson.kmi.open.ac.uk/Overview.html http://developer.yahoo.com/searchmonkey/ www.sti-innsbruck.at 78 78 References • Wikipedia links – http://en.wikipedia.org/wiki/Semantic_Web#Purpose – http://en.wikipedia.org/wiki/Semantic_Web www.sti-innsbruck.at 79 79 Questions? www.sti-innsbruck.at 80 80