Re: [ISSUE-77] [ISSUE-48] Re: [Dbpedia-discussion] Skos subject properties are deprecated

On 29/01/2008, Jakob Voss <jakob.voss@gbv.de> wrote:
>
> Dear Antoine and all,
>
>
> Antoine wrote:
>
> > I'm sorry but I don't fully understand your objections: what I call
> > "categorization"(or whatever the proper name be) in wikipedia is indeed
> > linked to an application, the one that will return a set of documents
> > that are relevant for the "history of the internet" each time I browse
> > this category. And the success of this application is dependent on the
> > way the link between documents and categories are related.
>
> In the Wikipedia case a set of documents is returned, in other cases it
> could be a set of people, a set of countries, or any other records.
> Frankly you cannot link to a person but only to a record that represents
> the person so in the end everything is a document - but not a document
> the the common understanding of "document".

In practical terms this is like saying that there is no recognition of
people, just of their drivers license. You don't refer to people as
just being a drivers license number, you refer to them, so you avoid
ambiguities for instance if they change countries and get a new
drivers license, but are still the same person. Wikipedia, I believe,
aims to describe categories based on these non-document-only
properties, where documents are representations, not the real thing.

> > I agree that the relation we are searching for connects a concept to a
> > resource, but I think it should have more semantics: otherwise you could
> > use it to represent a link between one concept and the resource standing
> > for its creator, who might have very little to do with this subject the
> > concept resource stands for. And I do think we can refer to a class of
> > document description and/or retrieval applications that share enough
> > commonalities to give a precise (or inprecise) enough idea of what these
> > semantics are.
>
> After reader your comments I'd prefer to call the relation
> skos:indexedWith, but I doubt that such changes in the naming of
> relations will be made to keep compatible with the previous draft.

It was a draft, and as such should have been taken by users to be
non-final, but in some way useful. If the vocabulary terms are
changing then it is incompatible by default IMO, subject to some user
based change, or recognition of a status quo by a user that their
intention with using the term has not changed.

My main objection with the subject term is that it is restricting its
universe to abstract concepts, which are always represented by
documents only, as they have no real world entity to claim to be
representing. Concepts work fine for the internal parts of a
classification scheme, but they don't work when you attempt to talk
about a non-abstract-concept being an actual subject.

Peter Ansell

Received on Tuesday, 29 January 2008 09:50:28 UTC