Definitions (was something else)
Alasdair J G Gray
agray at dcs.gla.ac.uk
Thu Feb 7 05:55:07 PST 2008
Rob Seaman wrote:
>> I would say that each vocabulary is meant to have its own definition.
>> This could be the same as another vocabularies definition for the
>> term, but it should be explicitly stated in each vocabulary.
> In that case, I'm concerned that we're suddenly:
> 1) signing on to generate thousands of definitions;
> 2) planning to extend other people's work without consulting them
> about what they really thought each term meant;
> 3) cross-linking terms asserted to be synonyms that may have subtly
> (or drastically) different definitions;
> 4) have established the need to link to external definitions, if only
> in other lists;
> 5) dramatically raised the bar for future participants.
> Can we make the definitions optional? ("zero or more", "0..*")
At the moment, the definitions are optional, and I do not think this
will change. (Certainly it will not change in the skos standard.)
I also agree that it is not our place to add definitions to vocabularies
published by external sources, e.g. A&A keywords or even the AOIM. If
the authors/creators/publishers of these vocabularies make definitions
available for the terms then these should be included, otherwise they
should be left as just the label and the relationships.
The only vocabulary within this standardisation process for which we
should worry about definitions is the new IVOAT.
> Can we instead collect a separate master list of numbered definitions
> that are linked to each entry, rather than embedded?
> In this way, the author(s) of a list can assert a definition and later
> users of the list can revise this via definitions that may already
> exist (and that may be shared with other lists). Adding a new
> definition is trivial. Stale definitions can be retired simply by
> replaced text in the master list with a link (just as real
> dictionaries might do, e.g., "see entry PLANTAIN" - in our case, "see
> entry 1234").
I do not favour this approach. Since we are only needing definitions for
one vocabulary, these should be placed directly into the vocabulary.
> May I also say that writing a thousand short descriptions may be many
> times the work of writing one coherent passage a thousand times as
> long. Ask any poet about the overhead of finding just the right
> words. (Or better yet, can we find an actual dictionary author to
> comment on our work products?)
> I am unpersuaded by the argument that any old paragraph will do to get
> the gist of each term. Either few will consult these definitions - in
> which case, why do the work? - or many will, and it should thus be the
> best the VO can produce.
> - Rob
Dr Alasdair J G Gray
Tel: +44 141 330 6292
Postal: Computing Science,
17 Lilybank Gardens,
University of Glasgow,
G12 8QQ, UK.
More information about the semantics