IVOA Thesaurus

[Frederic V. Hessman]

On 25 Oct 2007, at 4:42 pm, Alasdair Gray wrote:

> I have just been analysing your IVOA Thesaurus. (The version I have  
> used is Wed Oct 24 15:25:24 2007.)
>
> First of all I would like to talk about top level concepts. I found  
> that there were 2,646 top level concepts. I feel this very high,  
> particularly as there are only about 2,850 concepts in total. This  
> is also particularly bothersome if you analyse the IAU Thesaurus  
> which only has 516 top level concepts.
The immediate answer is, of course, that most of the structure is  
inherited from the IAU thesaurus.   This means that there are lots of  
superfluous NT's and RT's and missing BT' s and NT's.
> From this I have a question for everyone, how many top level  
> concepts does it make sense to have?
>
> My opinion about top level concepts is that they should aid the  
> user to explore the vocabulary. As such, I believe that there  
> should be only a limited number of top level concepts that can be  
> easily displayed at one time, e.g. at most 50 but probably  
> significantly less.

Ah, that was the idea in the original SV proposal, but we were  
collegially bashed on the head for that.   I'd say that it pays to  
clean up the hierarchy, but shouldn't the number of top-level  
concepts be determined by a clean set of BT's and NT's?.  One has as  
many as one has.

That said, we are not yet in possession of a clean set of BT's and  
NT's, so I'm hoping that Alasdair's set of errors/changes below will  
be a step towards this and there is hope that we can reduce the  
number of top-level tokens.

Note also that I'm using the high-level ontological tool called "vi"  
to maintain the text file versions.  After you've turned 50, you're  
allowed to be so old-fashioned, but I'd love somebody to turn this  
process into a Wiki-based one......   The RTML schema used to be  
developed that way - make a change in the entry, and the schema got  
changed, the diagrams got fixed, and the links were all made - like  
magic and all online (until the great student went away and I didn't  
have time to keep it up-to-date.......).
> In going through the concepts that appear at top level concepts, I  
> can see plenty of terms that appear as top level concepts that I  
> would think are narrower than other terms, e.g. I would think that  
> telescope mountings should be a narrower term than telescope.  
> Likewise for the many other concepts to do with different features  
> of telescopes.
By all means - the more corrections that are submitted, the better it  
gets!
> My second issue is to do with maintenance. In going through the  
> vocabulary I have found quite a few errors that have crept in due  
> to the editing process. (The list can be found at the bottom of  
> this email.) Some of these are simple typos but others are where  
> identifiers have been changed for a concept and the subsequent  
> edits have not been made for all of the relationships involving  
> that edit. Others make the vocabulary poly-hierarchy inconsistent,  
> e.g. acceleration of gravity has gravity as a broader term but the  
> inverse relation is not present in gravity. I think that we need to  
> develop scripts that can check that the inverses of all of the  
> relationships that are declared in the hierarchy are present. This  
> would be based on the related relationship being symmetrical and  
> the broader/narrower being inverses of each other.
I have made very little effort to correct the consistency of BT's and  
NT's, thinking like Alasdair, that someone will use a much fancier  
tool which spits out the correct list.

Note, however, that total consitency is only really important when  
the thesaurus is more-or-less finished in it's final working-draft form.
> I hope this all helps. Rick, please keep up the good work on  
> generating the IVOA thesaurus.

Thank you for actively pushing and shoving!  May it set a good example.

> Errors that I have identified in the thesaurus:
>
>
>
> The following are labels that I believe appear as identifiers in a  
> relationship but have not been defined as concepts in their own  
> right. This is probably due to typos or the changing of identifier  
> labels.

Yup - these have all been corrected.   Will upload the latest version  
of the IVOAT vocabulary on Monday (I'm changing the HTML to make it  
easier to browse).

Rick

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.ivoa.net/pipermail/semantics/attachments/20071026/69f97c45/attachment-0001.html>