From: Peter F. Patel-Schneider (pfps@research.bell-labs.com)
Date: 10/01/01
Here are my promised points on the proposed agenda items: 1/ Comments about the model theory itself, as it relates to RDF and RDF Schema, that we may want to get back to the RDF Core WG. a/ This was a very good job by Pat, even with the mostly minor points that I brought out last week. b/ The only remaining major point that I have is that the model theory commits RDF to a particular view of how literals obtain meaning. [See below as well.] In this view the complete meaning of a literals is obtained by its lexical form, which means: i) the same lexical form, i.e., 010, has to have the same meaning everywhere, i.e., can't be a string sometimes and an integer otherwise; and ii) the meaning of a literal cannot be (partially) defined by domain and range constraints. 2/ Discussion of the impact of the model theory on DAML+OIL. a/ The non-literal part of the model theory seems to me to fit very well with what we need for DAML+OIL. In partcular, I have modified the model theory for DAML+OIL to fit with the RDF model theory. (See my messages over the weekend.) In my view, this enhances DAML+OIL, and provides an interesting new direction for work on description logics, that of allowing a weak theory of types into the object language without messing up the rest of the description logic theory. (At least I hope so. I haven't completely thought out the consequences of this approach.) I would like to get more description logic theoreticians involved in this. b/ The literal part of the model theory is quite restrictive. It does not allow the meaning of a literal to be determined by RDF-supplied typing. I view this as a serious, possibly fatal, flaw, particularly with respect to DAML+OIL. 3/ Discussion of the political aspects of the situation with respect to the RDF Core WG, the WebOnt WG, W3C, and the Joint Committee. a/ I strongly feel that the delay in the formation of the WebOnt WG is hurting RDF efforts, as there is no formal mechanism for the concerns of the Joint Committee to get to the RDF Core WG, nor is there a good mechanism for information from the RDF Core WG to get to the Joint Committee. b/ In my view, the RDF Core WG is only hurting itself and RDF by moving further from XML Schema standards with respect to literals. Again in my view, they should be exploring mechanisms for bringing RDF literals in line with XML Schema datatypes, and not inventing a new, complex syntax. peter
This archive was generated by hypermail 2.1.4 : 04/02/02 EST