Joint Committee Minutes 29 May 2001

This page summarizes the telecon for the Joint US/EU Committee on Agent Markup Languages held from 1300 to 1400 PDT on 29 May 2001. These minutes were prepared by the chairman, and were approved during the 5 June telecon.



Thanks to Deb McGuinness for chairing the meetings on May 8 and 15.

Last Week's Minutes

The minutes from May 15 were approved by those present.

Because Mike was late in getting them out, we'll review the minutes from May 22 next week.

RDF Issues/Proposals for RDF Core WG

The W3C RDF Core Working Group is working primarily from the RDF Issues List. We've been asked to make sure that DAML issues are being addressed.

Mike found several messages containing DAML issues for RDF:

Additional RDF Issues relevant to DAML include

There was considerable discussion about reification. The term may be over-used in the RDF community. Pat sees no problem with data structuring (building more complex graph structure out of triples) but doesn't think that should be called reification. Quoting is a traditional way in logic of representing reification.

There was no consensus on whether we would prefer that reification be removed from RDF, but it was agreed that DAML+OIL has no dependency on reification

DAML-Rules Proposals

We began to discuss the rules proposals from Peter, Pat, and Stefan.

There was general agreement that use of rules will inevitably make the DAML language undecidable.

Peter's goal was to provide the minimal extension that would be useful. In particular, he only allows a single antecedent (analogous to guarded fragments or safety conditions on Horn clauses, Crawford and Allen's access-limited logic, and other attempts to make Horn clauses more computationally tractable).

There was considerable discussion of how the "class semantics" relate to the "rule semantics" and how this affects the complexity. Stefan suggested that by basing rules at the RDFS layer (rather than DAML+OIL), we might avoid the "hybrid semantics". In contrast, Dan would like to express some of the axiomatic semantics as rules. Ian suggested that these may be implementation issues rather than language definition issues. There was consensus that the DAML+OIL documentation should note that DAML doesn't require completeness (e.g. the ability of rules to affect classification).

Stefan referred people to CARIN, which combines Horn Rules and Description Logics.

Next Week

We'll plan to discuss Stefan's updated layers proposal and go into more detail on the above rules proposals.

ACTION (everyone): review the current DAML FAQ and be prepared to discuss fixes, additions, etc. before we make it public.


Kelly Barber's raw notes
last week's minutes
Joint Committee home page

$Id: 2001-05-29.html,v 1.6 2001/06/08 06:19:14 mdean Exp $