Transcript Document
The 2nd International Semantic Web Conference (ISWC03) 20-23 october 2003 Sanibel Island (Florida) Michele Missikoff Federica Schiappelli Summary The conference program Overview of the Semantic Integration workshop The main conference Keywords for the Semantic Web Roadmap (by Tim Berners Lee) 2 The conference “events” Tutorials Workshops Keynote speaches Panels Main conference Posters presentation Semantic web challenge http://iswc2003.semanticweb.org/ 3 Tutorials (on Monday, 20 oct.) Agent-Mediated Semantic Web/Grid Services Katia Sycara and Terry Payne Tutorial on OWL Peter F. Patel-Schneider, Ian Horrocks, and Sean Bechhofer Creating Ontologies and Semantic Web Applications with Protégé Holger Knublauch and Natasha F. Noy Information Integration on the World Wide Web Heiner Stuckenschmidt, Ubbo Visser, Holger Wache http://iswc2003.semanticweb.org/pdf/Protege-OWL-Tutorial-ISWC03.pdf http://webode.dia.fi.upm.es/iswc03/ 4 Workshops (on Monday, 20 oct.) 1. Practical and Scalable Semantic Systems 2. Semantic Integration 3. Semantic Web Technologies for Searching and Retrieving Scientific Data 4. Human Language Technology for the Semantic Web and Web Services 5. Rules and Rule Markup Languages for the Semantic Web 6. Evaluation of Ontology-Based Tools 5 Summary The conference program Overview of the Semantic Integration workshop The main conference Keywords for the Semantic Web Roadmap (by Tim Berners Lee) 6 The information integration problem 7 The Semantic Integration workshop In the Semantic Web context, the information are described by multiple ontologies and schemas Matching between ontologies and schemas is still largely done by hand Numerous research activities on methods for describing mappings, manipulating them, and generating them semi-automatically Electronic proceedings: http://ceur-ws.org/Vol-82/ Invited talks (slides): http://smi.stanford.edu/si2003/invitedTalksAbstracts.html 8 Keynote Speeches and Panels (on Monday, 20 oct., during the workshop) Semantic Web Scenarios involve rendez-vous between peers Requires mappings between their ontologies Generate mappings Translate between ontology languages Maintain mappings as ontologies change The same problems as database schema integration, BUT Current approaches to data management are not enough: language-specific; problem-specific Philip A. Bernstein Microsoft Research Generic Model Management: A Database Infrastructure for Schema Manipulation 9 A specific solution proposed by Bernstein A generic approach: model management operators to manipulate models and mappings as bulk objects A model is a rooted directed graph, which represents a complex information structure A mapping is a model that represents a transformation between two models… …Or it could be a binary table (a morphism) Schema matching (mapping discovery) Given two schemas, return correspondences that specify pairs of related elements (lexical-structural-superclasses alignement) Semantic Mapping Given correspondences between two schemas, return an expression that translates instances of one schema into instances of the other. Model Merging Use the mapping to guide the merging 10 Keynote Speeches and Panels (on Monday, 20 oct., during the workshop) – cont’d Two principal paths for info integration Using a central structure as ‘interlingua’ Problems: Creating the central structure (coverage, consistency, updating) Linking sources and targets to it (automatically?) Benefits: Linear (2N) in number of sources/targets Creating individual source-to-target mappings Problems: Creating and updating the mappings (automatically?) N2 in number of sources/targets Benefits: Doesn’t require general one-size-fits-all model/structure Edward Hovy Information Sciences Institute of the University of Southern California Building Large Ontologies 11 Hovy’s approach The Interlingua route: toward a merged ontology General alignment and merging problem Given many domain models—how can you integrate them consistently, without overlap or redundancy? Solution: Use a large general-purpose concept network to provide the background—the SENSUS Ontology. Improving alignment by enriching content by adding definitional material and by clustering entities. The Transfer route: toward learning individual mappings aligning databases directly 12 Summary The conference program Overview of the Semantic Integration workshop The main conference Keywords for the Semantic Web Roadmap (by Tim Berners Lee) 13 Invited Speakers (during the main conference) Jim Hendler: On Beyond Ontology: Returning to AI from the Semantic Web Michael Brodie: The Long and Winding Road To Industrial Strength Semantic Web Services Tim Berners-Lee: SemanticWeb:Where to direct our energy? 14 Jim Hendler’s talk Director of a University of Maryland's lab; cochair of the W3C Web Ontology Working Group his research group developed SHOE; creator of DARPA's DAML program “…it's beginning to look like we may be successful! ” OWL, the Web recommended ontology language Logic toolkits are produced by software companies Time for us to think more about what we do with all this Today Challenges: Integration… partial mapping Process modelling … WSBPEL, OGSA, etc. Temporal logic OWL feeding Common agreement for ontology building Semantic Web technology available to vendors Electronic proceedings: http://ceur-ws.org/Vol-82/ Invited talks (slides): http://smi.stanford.edu/si2003/invitedTalksAbstracts.html 15 Michael Brodie’s talk Chief Scientist, Verizon Information Technology industrial researcher, focussing on advanced computational models and architectures, the large-scale information systems that they support, business and technical contexts “Web Services: the basis for the Next Generation of computing ! ” flexible can be discovered and invoked anywhere composed, as required, to achieve higher level goals proposed to address software integration Today Challenges: overcome the integration challenge on an industrial scale technical pragmatics such as scalability and performance dominate http://iswc2003.semanticweb.org/invitedtalks.html 16 Tim Berners-Lee’s talk The Semantic Web inventor! Speaking too fast… didn’t understand anything… See the conclusions!!! 17 The main conference Principal themes Foundations Ontological reasoning Semantic web services Security, trust and privacy Agents and the Semantic Web Information Retrieval Multi-media Tools and methodologies Applications Industrial Track 18 Summary The conference program Overview of the Semantic Integration workshop The main conference Keywords for the Semantic Web Roadmap (by Tim Berners Lee) 19 Interesting themes Interoperability Contexts Languages: RDF(S); OWL Reasoning with DL Web services…composition? 20 Interoperability Semantic coordination (solution by Trento univ.) Semantic models considered are hierarchical classifications, represented as labelled graphs Logical formulae are built taking into consideration lexical knowledge (words in labels), domain knowledge (relations bw concepts represented by labels), structural knowledge (isa hierarchy). Shift from computing linguistic and structural similarity to the problem of deducing relations bw sets of logical formulae, encoding the meaning of the involved entities (nodes on the graph) P. Bouquet University of Trento Semantic Coordination 21 Contexts Context is a model of some domain, which encode a particular view Context is local (reduced sharability) Mapping among contexts is the issue Contextual Ontologies = Ontology + context mapping Fausto Giunchiglia University of Trento C-OWL: Contextualizing Ontologies 22 Languages: RDF(S) RDF(S) has a non standard metamodelling architecture Multiple modelling primitives seem to be represented by the same RDFS primitive (e.g. rdf:type, rdf:subClassOf) A Fixed meta-modelling architecture has been proposed I.Horrocks F.Patel-Schneider University of Manchester RDFS(FA) and RDF MT: two semantics for RDF 23 A non standard meta-modelling architecture RDF(S) is used to add metadata annotations to Web res. Subject-predicate-object triples used to link resources i.e., triples represent knowledge about domain (such as Federica worksWith Francesco) Federica Francesco worksWith RDF(S) also used to define syntax and semantics of subsequent language layers (and even of itself), e.g.: Parent Restriction hasChild subClassOf equivalentClass onProperty subClassOf subPropertyOf minCardinality 1 subClassOf Class Resource type 24 Problems with RDF MT Not clear that RDF(S) is appropriate for both functions (at once) Uniform semantic treatment of triple syntax i.e., “syntax” and “knowledge” triples have same semantics I use Confusing (for some) cyclical Should meta-model owl:Class or rdfs:Class? Semantics given by “non-standard” Model Theory instance of rdfs:Class More expressiverdfs:Resource ontology languages layered rdfs:Class subclass of rdfs:Resource …Resource is instance of its subclass?? on top of RDF(S) E.g., OIL, DAML+OIL, and now OWL 25 RDFS(FA) RDFS(FA) is a sub-language of RDF(S) It stands for “RDFS with Fixed layer metamodeling Architecture” Has a First Order/Description Logic style semantics The universe of discourse is divided up into a series of strata User defined facts, vocabulary and RDF/OWL builtin vocabulary are (typically) in different strata Each modelling primitive belongs to a certain stratum (layer) Labelled with different prefix to indicate the stratum 26 RDFS(FA) layers Stratum 3 (Meta-Language Layer) fa:MResource, fa:MClass fa:MProperty … Stratum 2 (Language Layer) fa:LResource, fa:LClass fa:LProperty … Stratum 1 (Ontology Layer) fa:OResource Person, Researcher workWith … Stratum 0 (Instance Layer) Federica, Francesco… 27 Advantages of RDFS(FA) No problems layering FO languages on top of RDFS(FA) RDFS(FA) supports use of meta-classes and meta-properties In stratum above classes and properties RDFS(FA) metamodel very similar to that of UML Possible to define a new sub-language of OWL: OWL FA Extends OWL DL with meta-classes/properties Fully compatible with OWL DL semantics Reasoning (even for meta-classes/properties) 28 Reasoning with DL Reasoning with ontology languages is important to exploit the semantics of ontologybased annotations Instance checking Subsumption (taxonomic) reasoning Used in SW applications E.g. search engines, matchmaking of services, document classification, etc… OWL is strictly related to Description Logics DL provides such reasoning facilities I.Horrocks F.Patel-Schneider University of Manchester Reducing OWLentailment to DL satisfiability 29 Web services (?) Karlsruhe: WS Composition is a planning problem or pre-/post-cond matching OntoMat-Service (tool for WS workflow) S.Agarwal, S.Handschuluh, S.Staab University of Karlsruhe Surfing the Service Web BPEL4WS (Stanford Univ.) Coreography Fwk BPEL (programming lang) to specify the sequence of tasks Partner selected at runtime Automatic semantic translation D.J.Mandell, S.McIlraith University of Stanformd Adapting BPEL4WS for teh SW: the bottom-up approach to ws interoperation 30 Summary The conference program Overview of the Semantic Integration workshop The main conference Keywords for the Semantic Web Roadmap (by Tim Berners Lee) 31 SW status OWL becomes stable Steadily growing deployment of RDF Growing SWeb-specific industry sector SW Services starting to take off 32 Risks Architecture becomes fractured, weak, or baroque Fracture between Web and S/Web arch Fragmentation in query and rules RDF/XML syntax shock Perceived relationships between SW and WS Deployment in real products 33 The Killer App for the Semantic Web It’s the integration!!! Guidelines Be careful of terms - ontology, semantics, etc Explaining how communities interact Please re-use Don't create new URI schemes Don't re-invent HTTP space Don't re-invent RDF Don't re-invent ontologies where they exist 34 Where to direct our energy? Indexing data - by ontology Indexing rules, building translation paths like one big database? or one big web? SW and WS Discovery should be SemWeb-based Balances Engineering vs Research Getting it working vs getting it right Tractable Machinery vs Heuristics 35 Thank you for the attention Mmm mm 36