- From: Peter Ansell <ansell.peter@gmail.com>
- Date: Wed, 30 Jan 2008 06:21:44 +1000
- To: SKOS <public-esw-thes@w3.org>
- Cc: "Simon Spero" <ses@unc.edu>, "Antoine Isaac" <Antoine.Isaac@kb.nl>
On 28/01/2008, Antoine Isaac <Antoine.Isaac@kb.nl> wrote: > Dear Simon, > > All your quotations make sense to me. > The problem with 5 is that in an open world as the semantic web, we can > define the extension of a concept, but it's much more difficult to access > (at least more than in traditional library databases where, despite the > acquisition of new books, you can still access *all* the documents about a > given subject). And if several sources use a same concept they might use > different flavors for the "indexing" link, or one could just produce crap > indexes... > I can understand the reluctance of some people to have SKOS introduce > skos:subject on the grouind that it will define the extension of concepts. > In practice, it would just help to access a (maybe debatable) part of its > extension. Which is still better than nothing imho. I would regard skos:subject as being a link out from the knowledge scheme to indicate the place of the knowledge scheme in the overall scheme of things (ie, the open semantic web). Linking out of the locally defined scheme infers that you are referencing any resource using the subject identifier, as you can't rely on the fact that everything will be a conceptual abstract unit of thought in the open web where Resources essentially form the basic unit of information which can be identified. I see it as reasonable to define a way to relate skos to the real world (or at least show your intention by avoiding the intentionally abstract skos:concept definition), as this would help in cases where that is actually what you want to do. The essential parts of the index are contained inside of the skos vocabulary, with external references being a use-at-your-own risk type area. I don't think it would produce indexes of lower quality than if someone was forced to redescribe each concept under some other wrapper, or alternatively link to it using the generic seeAlso link which is use-at-your-own-risk anyway. The only advantage I see in skos:subject over seeAlso in this scenario is that it allows one to say what the idea was with the link using the locally defined vocabulary term. Subject to me essentially describes an equal level link, as in it is not a category or a term, just something that is similar in knowledge value, but which is contained outside of the local vocabulary. As Simon referred to, this allows all content objects to be useful in placing the concept scheme in a bigger scheme. Peter Ansell > -------- Message d'origine-------- > De: public-esw-thes-request@w3.org de la part de Simon Spero > Date: lun. 28/01/2008 00:53 > À: SKOS > Objet : Are the following statements valid? > > > [I just want to check if the following assumptions are generally > acceptible . If 5 is ok, it makes some arguments easier ] > > A Concept in SKOS represents "a unit of thought [..] combining some or > all of the characteristics of a concrete or abstract, real or > imaginary object." (Z39.19, p. 4) > Concepts are represented in textual forms by indexing terms. > (Z39.19, p. 6). > indexing is the process of assigning indexing terms to content objects > (Z39.19 p6). > Content objects are things which convey information/data. (Z39.19, p. > 4). > The extension of a concept is the set of all content objects which > (are? could be?) indexed with indexing terms corresponding to that > concept. >
Received on Tuesday, 29 January 2008 20:21:55 UTC