Joint Committee Minutes 9 October 2001

This page summarizes the telecon for the Joint US/EU Committee on Agent Markup Languages held from 1300 to 1400 PDT on 9 October 2001. These minutes were prepared by the chairman, and were approved during the 23 October telecon.

Participants

Announcements

No announcements this week.

Last Week's Minutes

The minutes from October 2 were approved by those present.

Semantic Web Course Materials

Several folks offered to share (with attribution) materials they've developed for Semantic Web courses and industrial presentations Frank also mentioned that Arthur Stutt is collecting educational resources as part of the OntoWeb network.

NSF/EU Semantic Web Workshop

Frank, Deb, Stefan and others reported on the workshop held last week in Nice, France. The goal was for current researchers to brief the government representatives on the research agendas should be. There were about ~15 presenters (several US presenters cancelled due to concerns about travel).

NSF is interesting starting a Semantic Web activity but the representative there was very new. They are likely to organize a follow-on workshop.

The EU representatives (the folks who sponsored OntoWeb and other efforts) are currently writing a larger "framework" for the next 5-year period (2003-2008).

After the presentations, the group divided into 4 working groups:

Model Theory

Pat and Peter discussed their recent work on the RDF(S) and DAML+OIL Model Theories, for those that weren't present last week. Pat noted that the current Working Draft will be replaced shortly, pending decisions regarding literals.

Frank complimented Pat on writing an "eminently readable document" and noted that we should produce more documents like this.

Peter identified 2 issues in layering the DAML+OIL MT over the RDF(S) model theory:

  1. the DAML+OIL disjointness of objects and datatypes (due to undecidability) is hard to fit with RDF Resources and Literals
  2. the RDF(S) MT employs a fixed mapping from Literals to literal values, while DAML+OIL can map a literal value into any of the types it could match

Dan Connolly noted that the current DAML+OIL documents don't contain the disclaimer expressing our discomfort with the disjointness of objects and datatypes (but it's the best we currently know how to do) that we had planned to add.

RDF(S) Data Types

Peter discussed his recent work (here and here) on RDF(S) datatypes. He plans to send out a message shortly. He was originally looking for Joint Committee support, but is now comfortable proposing as an individual.

As noted above, a DAML+OIL "literal" can match any of the types it could match, and may be contrained by a subsequent restriction. In RDF(S), it has only one meaning.

Peter's new Model Theory is slightly cleaner than the previous DAML+OIL Model Theory, excising some obfuscation with XML Schema datatypes that are hard to reason about (e.g. multiple numeric types), but is compatible with the previous DAML+OIL Model Theory.

XML vs. RDF

There was some discussion about the (changing) relationship between XML and RDF, which has come up at the EU/NSF meeting and elsewhere.

Originally RDFS was intended for domain models, while XML Schema was intended for document models. These are now converging.

Dan Connolly fees that the biggest difference between RDF and XML is the inability to merge XML documents.

Ian suggested the name "XML Ont", and Deb suggested that this should be the name for the product of the proposed W3C WebOnt WG.

Peter and Jerome Simeon (a Lucent colleague "across the hall" working on XML Query) are looking at a merge of XML and RDF. Peter noted that the XML Query and XPath Data Model is much richer than RDF, including ordering, etc.

Harold Boley is also working on unifying data models (where XML is a special case of RDF).

Pat Hayes is also working on a unified logical notation with a common semantics/model theory that can project to XML, RDF, KIF, etc.

Next Week

Several members will be en route to the DARPA Rapid Knowledge Formation (RKF) PI Meeting next week. We'll plan to have at least a short telecon, anyway, particularly in case there is any news on WebOnt.

On October 23, we'll plan to discuss extra-query information.

During some future telecon, we'll discuss the XML Query Data Model. ACTION (DanC): send information on the XML Query Data Model to the group

Links

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

$Id: 2001-10-09.html,v 1.10 2001/10/24 20:15:35 mdean Exp $