- From: Andrei S. Lopatenko <andrei@derpi.tuwien.ac.at>
- Date: Wed, 22 Aug 2001 17:21:53 +0200
- To: "Peter F. Patel-Schneider" <pfps@research.bell-labs.com>, "Leo Obrst" <lobrst@mitre.org>, <lacher@db.stanford.edu>, <gdm@empolis.co.uk>
- Cc: <www-rdf-interest@w3.org>
Yes, but "object layer" transformation is a good example that transformation is possible without information losses I think that "object layer" transformation could be a first step for semantic transformation, because it seems not hard (?) to extract explicit semantics from new object-layer RDF graph and put it into explicit semantic declaration (RDFS, DAML) and transform object-layer RDF graph into "semantic" RDF file according new schema definition Best regards MSc Andrei S. Lopatenko Researcher Vienna University of Technology Extension Centre http://derpi.tuwien.ac.at/~andrei/ ----- Original Message ----- From: "Peter F. Patel-Schneider" <pfps@research.bell-labs.com> To: <andrei@derpi.tuwien.ac.at> Cc: <lobrst@mitre.org>; <lacher@db.stanford.edu>; <gdm@empolis.co.uk>; <www-rdf-interest@w3.org> Sent: Wednesday, August 22, 2001 4:35 PM Subject: Re: On the integration of Topic Maps and RDF > Agreed, the information can be regenerated from the transformed RDF. > However, the regeneration cannot be performed in RDF, as some external > information is needed, whether in the form of some unspecified rules > language or in the form of informal side-agreements. > > My view is that semantic transformations should not require any of this > extra information. > > > From: "Andrei S. Lopatenko" <andrei@derpi.tuwien.ac.at> > Subject: Re: On the integration of Topic Maps and RDF > Date: Wed, 22 Aug 2001 15:50:32 +0200 > > > I completely agree that semantic information must not be lost in > > transformation. > > But really it was not lost in "object layer" mapping suggested in the > > article. > > Semantic information about meaning of RDF resources transformed from XTM is > > implicitly declared in RDF graph of resourses > > and could be extracted by inference engine. In F-Logic query example > > semantic of element is extracted in a such way (roleLabel condition). > > I would not state that the extra information is implicit in the RDF. It > requires extra information, such as inference rules, to be recovered. > > > So there are several possible ways to map TM into RDF > > > > 1 The one is preserving explicit semantic > > XTM -> RDF Resource graph + RDF Schema (or DAML+ OIL, OIL) > > For example, such classes as country, natural-resource should be defined in > > the schema > > And then query should be asked using that new terms > > ... > > natural-resource -> pertoleum; > > ... > > 2 Another is "object layer" mapping which just encode XTM graph as a RDF > > graph and semantic is stored implicitly in that graph. > > The query should contain statements for extracting semantic information > > such in F-Logic in the article > > ... > > tms:roleLabel->natural-resource; > > ... > > or semantic should be provided by inference engine > > > > But from the point of view of the article - to develop query engine which > > can also include XTM resources into RDF > > both ways are suitable. The difference is only in queries. Information is > > not lost. > > Not lost to a fully capable reasoner, such as a human. However, it is lost > to an RDF (only)-capable agent. > > > Maybe, for other applications it is neccesity to have explicitly declared > > semantic, bot not for this? > > > > Best regards > > MSc Andrei S. Lopatenko > > Researcher > > Vienna University of Technology > > Extension Centre > > http://derpi.tuwien.ac.at/~andrei/ > > Peter F. Patel-Schneider > Bell Labs Research > >
Received on Wednesday, 22 August 2001 11:16:52 UTC