Presentation on theme: "A rule language for the semantic web Dieter Fensel, Lausanne, June 14, 2004 SDK cluster meeting on WSMO."— Presentation transcript:
A rule language for the semantic web Dieter Fensel, Lausanne, June 14, 2004 SDK cluster meeting on WSMO
2 Language Penal This panel is about the discussion on the further development of OWL OWL is not the end-point of the semantic web OWL will be extended / complemented by languages that employ the potential of the Horn logic fragment of first order logic.
3 An example for this: SWRL SWRL: A Semantic Web Rule Language Combining OWL and RuleML W3C Member Submission 21 May 2004 This version: Authors: Ian Horrocks, Network Inference Peter F. Patel-Schneider, Bell Labs Research, Lucent Technologies Harold Boley, National Research Council of Canada Said Tabet, Macgregor, Inc. Benjamin Grosof, Sloan School of Management, MIT Mike Dean, BBN Technologies
4 Language Panelist We have the pleasure to have four world-leading experts here to discuss issues related to this. Guus Schreiber, expert on conceptual modeling languages Ian Horrocks expert on Description Logic Michael Kiefer expert on rule languages Dieter Fensel (panel chair) expert on scientific marketing
5 Six opinions of Dieter Fensel DL is an interesting subset of 1st order logic HL is an interesting subset of 1st order logic Full 1st order logic is an interesting language Pure logics are cumbersome tools for modeling SWRL is crap A useful rule language for the semantic web must follow a DL-minimalistic approach
6 (1) DL is interesting because of its decidability/tractability properties spoken for the man on the street it is the fragment of logic that can be computationally explored around the existential quantifier. it is therefore worth to become standardized in OWL besides the fact that I do not like at all OWL-Lite which is an overkill, a smooth extension of RDFS would have been much more appropriate.
7 (2) HL is interesting because of its decidability/tractability properties Spoken for the man on the street it is the fragment of logic that can be computationally explored around the all quantifier without function symbols it is decidable and with minimal-model semantics function symbols, non-monotinicity of negation, and even transitive closure (an extension of first order logic) can be expressed. It is therefore worth to become standardized by W3C, however, not in a way that destroy all features of rule languages by mixing it up too much with DL type of expressions.
8 (3) First order logic is interesting In oposition to DLs and HLs it provides significant more flexibility in writing down required axioms. This language is not fully mechanizable in terms of reasoning support, however, many interesting theorems can be proven and the theorem prover community has made significant progress over the last years. First order language could define the common umbrella, where DLs and HLs are sublanguages and unified.
9 (4) Pure logics are a cumbersome tool for modeling Therefore, extensions like Flogic that allow quantification over classes and attributes are that helpful. W3C recommendations should care for this if they want to make their languages widely adopted.
10 (5) I strongly dislike SWRL since it spoils the nice properties of rule languages for the price of defining the rule language as an extension of DLs. SWRL is meaningless since it defines a syntactical restriction of 1st order logic without any computational justification. SWRL is as undecidable and untractable as first order logic. Therefore, it is meaningless to syntactically restrict first order logic in that way.
11 (6) A useful rule language for the semantic web must follow a different approach It should emply the full power of the HL fragment and include DL features only in case they do not harm this language type. Therefore, a HL language should be defined as an extension of RDFS and/or OWL-Lite (where OWL Lite is reduced to an actual lite subset of DL).