- From: Dan Brickley <danbri@w3.org>
- Date: Tue, 6 Nov 2001 04:33:51 -0500 (EST)
- To: Pat Hayes <phayes@ai.uwf.edu>
- cc: <Patrick.Stickler@nokia.com>, <w3c-rdfcore-wg@w3.org>
On Mon, 5 Nov 2001, Pat Hayes wrote: > Ah, I think I see why we misunderstood one another. You are using > 'datatype' to refer to the set of things, and I am using it to refer > to lexical-to-value mapping. > > However, let me take up your point directly. We have a distinction > available between a class as an intensional object, and the set of > things in its extension. Somewhat as an aside, it occurs to me that one characteristic of 'datatype classes' not shared by all other rdfs:Class-es is that their extension is fixed, common across all interpretations. Classes such as wordnet:Person, rss:Channel, perhaps even rdf:Property have extensions that (to pick on the temporaral aspect of variation) will often differ through time. I have in mind that some extension of WebOnt/DAML+OIL will someday be able to make such observations mechanically evident, so that SemWeb apps can make braver inferences to support Web data merging. The notion that we called 'class sealing' in the RDF Schema 1.0 WG (but postponed to future work :) touched on this. It felt like a can of worms at the time, and probably still is, yet I wonder whether knowing that a class is 'static' in this way might be a useful distinction between 'datatype' classes and other species of RDF class. (I'm trying to think of the rdfs:label and rdfs:comment that one would give for an rdfs:Class of things that are datatype classes). Dan -- mailto:danbri@w3.org http://www.w3.org/People/DanBri/
Received on Tuesday, 6 November 2001 04:33:57 UTC