- From: Simon Spero <ses@unc.edu>
- Date: Mon, 17 Dec 2007 17:55:45 -0500
- To: "Daniel Rubin" <rubin@med.stanford.edu>
- Cc: "Sini, Margherita (KCEW)" <Margherita.Sini@fao.org>, "Ed Summers" <ehs@pobox.com>, public-swd-wg@w3.org, public-esw-thes@w3.org
On Dec 17, 2007 4:22 PM, Daniel Rubin <rubin@med.stanford.edu> wrote: > Sorry, but I disagree with the suggestion that people define > transitivity for themselves. SKOS should not leave such things > undefined--to do so guarantees people will have different semantics > for SKOS properties which will prevent interoperability. I agree with Daniel here. If broader is always transitive, the semantics are obvious, aligned with Z39.19, and close to subsumption. If broader *might* be intransitive, then any valid broader relations that could otherwise have been inferred must be explicitly included in the source vocabulary. I would much rather see broader/narrower remain always transitive, with different relationships for intransitive cases. An alternative would be to add cancel links (not-broader), and define a non-monotonic semantics for SKOS (see e.g. Morgenstern, 1997) . This approach is probably out of skope. Does anybody have a set of examples of valid relationships which they believe should be covered by 'broader', but which can't be treated as transitive? What are the semantics? What are the characteristics that distinguish these broaders from the transitive kind? Is there an ontologist in the house? Simon [Morgenstern, 1997] Morgenstern, L. (1997). Inheritance comes of age: Applying nonmonotonic techniques to problems in industry. In Pollack, M., editor, Proceedings of the Fifteenth International Joint Conference on Artificial Intelligence, San Francisco. Morgan Kaufmann.
Received on Monday, 17 December 2007 22:55:58 UTC