Format of tokens (was Re: Fwd: Re: IVOA Thesaurus)

[Ed Shaya]

Alasdair Gray wrote:
> > On Thursday 01 November 2007 6:32 pm Ed Shaya wrote:
>
> > Well, I vote to put back the underscores and the capitalization 
> where appropriate.  There is no need to go out of one's way and make 
> all IDs cryptic just to make a point about the concept of tokens.  In 
> ontology these become the element names of instances and it is really 
> handy to be able to readily discern what kind of instance it is by 
> looking, rather than going to some lookup table.  We need some 
> prescience here, not to be confused with pre_science.
>
> I would like to address Ed's comment about ontologies first and their 
> difference to vocabularies modelled in skos. In the skos world of 
> vocabularies, a user is not meant to use or see the ids. The 
> application that they use should display the preferred label which all 
> concepts should have (in all our examples they do have preferred 
> labels). A concept without a preferred label does not really make 
> sense either in a skos model or the real world as humans will use some 
> language phrase to describe the concept. Thus, the idea that the ids 
> become the element names of instances does not follow for vocabularies.
>
But  the preferred labels cannot be used by the ontology because they 
can have characters that are unacceptable as element names, notably 
spaces.  So if neither the label nor the ID in SKOS are acceptable for 
the ontology, then the ontology must have yet a third set of names and 
then there is no tie between the skos and the ontology.  I suppose one 
possibility is to add software that swaps the underscore with a space 
and has rules for other characters in order to compare to the label in 
skos and vice versa.  Or we can put underscores into the skos labels.  
But, believe me, you do not want to have an ontology where it takes even 
2 or 3 seconds for a human to decode each element name.

As for capitalization, one could probably get by without capitals, but 
a_star instance is already trouble.


Ed