- From: Simon Spero <sesuncedu@gmail.com>
- Date: Sun, 8 Jun 2008 12:44:21 -0400
- To: Antoine Isaac <Antoine.Isaac@KB.nl>
- Cc: SKOS <public-esw-thes@w3.org>
- Message-Id: <BD826FAA-E56E-4F84-9A21-819AFDD2EE51@gmail.com>
On Jun 8, 2008, at 9:53 AM, Antoine Isaac wrote: > To add to a previous mail of Leonard, I don't think we disagree on > what the 'real' indexing link should be for the document. > My problem is that if you add the axiom Alistair proposes, then a > knowledge base that has both SKOS information and dc:subject would > return, in a set of indistinguishable results, both the dc:subject > 'at the right level' and the more general subjects that can be > inferred from it. > Or more precisely: you can make the distinction, but then your query > to the knowledge base should be different, more explicit, like 'find > all the subjects of the book for which there is no specialization > which is also > The trouble with this approach is that it attempts to impose implementation of a reasoning system directly into the semantics of a knowledge base. There is no requirement for a specific inferencing process to compute every possible inference implied by a knowledge base; in fact, in many cases that's exactly what you don't want to do. This is all a function of the search/inference engine. For example, if 25% of all documents about animals are documents about cats, one might weight a document about animals not explicitly indexed with cats proportionately lower when broadening upwards. Again, this is all a function of the chosen inference engine, not the semantics of narrower and broader. Simon
Received on Sunday, 8 June 2008 16:45:12 UTC