Re: FW: URI policy for thesaurus concepts

On Mon, 3 May 2004, Stella Dextre Clarke wrote:

>
>While I agree with the desirability of enabling all these browse options
>(and also the option of downloading a whole thesaurus *only*
>occasionally) I don't follow the last para below, which seems to suggest
>that one term may at the same time be a preferred term for one concept
>and a non-preferred term for another. If the thesaurus conforms with ISO
>2788, then each term must be unique, whether preferred or non-preferred.

One of the points is the if.  If everybody followed all applicable standards
all the time many of us would be out of jobs. As you note below, there are
many real-world examples where people don't follow 2788, and some of these
arise in use cases such as the first attempt at merging two thesauri, or in
producing linguistic thesauri (like Roget's), or from time to time in
collaborative and dynamic development of a thesaurus.

My actual poinnt was that it would be interesting to have a note explaining
why a particular term was not preferred. This can be done with rdfs:commment
- for example, where developing a glossary for use in W3C specifications it
would be helpful to note for other developers that I avoided the term
"relative unit" as the preferred term for a concept because it was used by
someone else to describe a different concept. Nonetheless, the term is often
used to describe the concept I mean, and for a while it was a preferred term
(before it had to be used in a larger community, where the clash became
apparent). So it is helpful if people can search via the term and find both
concepts.

cheers

Chaals

>If two concepts in the same thesaurus could be described by the same
>natural-language word, then thesaurus terms may be concocted by adding
>qualifiers in parentheses, e.g. "bridges (electrical)" "bridges (roads)"
>"bridges (teeth)" etc. Any of these could be made either a preferred
>term or a non-preferred term, depending on the needs. But other forms of
>controlled vocabulary do not always follow the same rule.
>Stella
>
>One thing that is intersting in implementations is being abel to see why
>a certain term is not the preferred term for a concept - erhaps because
>it is a preferred term for another concept, or becausee it overlaps too
>much with its use as an alternative term for another concept...

Received on Monday, 3 May 2004 06:47:58 UTC