Using the Q model for the RDFTM guidelines

The RDFTM guidelines will need to include a section which describes mapping of RDF data to Topic Maps, and vice versa. This section will necessarily make up the most important content in the guidelines. The question is, how is the mapping best described? What formalism, if any, should we use?

The most obvious choices to describe the mapping are:

  • No formalism. Just use prose to describe the mapping. All proposals reviewed in the survey take this approach.

  • Use the Q model. This would give an approach pretty much as described in the Q paper

  • Find some other formalism. At the moment there are no obvious alternatives.

So, what do we prefer?

-- LarsMarius - 14 Sep 2005

Two main issues to start on the discussion.

  • Patel-Shneider had some objections in general about Q. We need to be sure to answer satisfactorily to these objections before we can consider Q as a valid candidate. Satisfactorily means two different and important things:
    1. Patel-Shneider is wrong in his opinions about Q and
    2. We can satisfactorly convince everybody else that he is wrong. This item does not necessarily follow from item 1.
  • It is not true that there were no obvious alternatives to Q. Valentina came out with a different formalism that had the advantage of being known and discussed in literature in the past. I am not ready to abandon this formalism. What was it, Valentina?

-- FabioVitali - 14 Sep 2005

The possible alternative we discussed in Oslo is the "Ontology definition Metamodel" !(ODM). The latest version of the ODM specification is available at I didn't abandon it, but I'm not sure about its usability. I propose to take it into account and ask the WG for opinion in the same post I'm going to ask about teir position on the Q model. Anyway I need to study deeper the way to use ODM for our purpose. I also propose to move this discussion on the list once I've done the post (possibly tomorrow) smile

-- ValentinaPresutti - 14 Sep 2005

I'm happy to look at other formalisms, provided we can get them on the table to be discussed.

As for Patel-Schneider's "objections" they can be found in the mail list archive. I think they could be summarized as follows:

  • The Q model only unified the "abstract syntaxes" (I'd call this data models, but never mind), and did not propose a combined semantics.
  • It is not obvious that a combined semantics will be workable, given that Topic Maps have things like scope etc.

My answer to this (also posted to the list) was that unification of the syntaxes is already a huge improvement over where we were, and an outline of how I think a unified semantics can be created. (Though admittedly it may still fail.) Patel-Schneider never replied.

My gut feeling is that the whole thing can be summarized as Patel-Schneider not thinking "syntax unification" is much to be excited about, and also thinking that I'm claiming to have solved the problem when to him a solution that doesn't handle the semantics is not a solution at all. That's my theory, anyway.

-- LarsMarius - 15 Sep 2005

Revision: r1.5 - 15 Sep 2005 - 10:35 - FabioVitali
Copyright © 1999-2018 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding Fabio's Wiki? Send feedback