From: Harold Boley ([email protected])
Date: 03/18/03
Hi JC Colleagues, Peter started today's discussion on using OWL in its favorite (perhaps XML-based) syntax and using subClassOf of RDFS for encoding the DL-generated class hierarchy in a second document type. Both can be regarded as 'terminological documents' (T-Docs, webized T-Boxes). As we had discussed earlier (http://www.dfki.uni-kl.de/ruleml/rdf/ruleml-rdf.pdf), RuleML derivation rules (e.g., webized Horn rules) can then access such an RDFS class hierarchy, via urirefs, for variable typing. Similarly to a DL classifier generating an extensional class lattice from a set of intensional class definitions, a (bottom-up/forward) rule engine generates instance descriptions / facts via derivation rules. Both facts and rules can be regarded as 'assertional documents' (A-Docs, webized A-Boxes) [although rules may also be viewed as separate 'inferentional documents' (I-Docs, webized I-Boxes)]. A second link between the T-Doc and A-Doc layers is urirefs pointing from constants (often URIs) in instance descriptions / facts 'up to' the extensional class lattice, e.g. via rdf:type. Finally, Benjamin's and Ian's description logic programs explore direct connections between derivation rules and intensional class definitions. In summary, we get this picture: T-Docs: Intensional Class Definitions Extensional Class Lattice (in OWL, queried by DQL) --generate--> (in RDFS) ^ ^ ^ | | | Description | | | | urirefs for typing variables | | urirefs for Logic | | | typing constants | --------------------------------- | Programs | | | | | | | | | A-Docs: v | | Derivation Rules Instance Descriptions / Facts (in RuleML) ---------------generate--> (in OWL, RDF, RuleML, DBs, ...) A version of this picture might help us keeping together the different KR communities and 'cultures' that have been developing languages for the Semantic Web. Best, Harold
This archive was generated by hypermail 2.1.4 : 03/18/03 EST