Re: SKOS/ synonym provenance

Hi Daniel, Alan,

You are actually making points in favor of this many-to-one option, 
which will be interesting to remind if we go for the label-as-resource 
option (which, I remind you, is not the current choice of the WG!)

I wonder however how your specific case can fit the SKOS world: what you 
call synonym here could be a case of skos:altLabel between two 
concepts in different communities and the "study" label, couldn't it?
Also, I don't think SKOS should propose means to represent provenance 
from entities different from concept schemes. Your "communities" seem a 
very specific requirement. And the problem is difficult enough for 
concept schemes, I'd say :-(

But at least we can try to have a basis that fits your representation 
needs in a reasonable way. It would 
therefore help if you could say whether you prefer represent your 
information using label-as-resources or Alistair's n-ary patterns for 
labels-as-literals [2]. Every user's advice is welcome on this point.




> Hi everyone,
> Alan Ruttenberg is interested ability of SKOS to support associating 
> information with synonyms, such as provenance, or recording the 
> community that uses the term. He had some comments he wanted me to 
> share with SWD on our issue called Label Relations under active 
> discussion [1]. Please see his comments below.
> Daniel
> [1] http:// 
> relations.html
> ______________
>> I think we would say that we need the many to one
>> case, as if the literal was inverse functional we would not be able
>> to have the same word be a synonym for different terms for different
>> communities.
>> So take the case of the term "study".  In the clinical trial
>> community this would be a synonym for ClinicalTrial, but in the
>> Nutrigenomics community it is a synonym for some portion of a
>> clinical investigation.
>> Therefore our case resembles the cow case, but is more clearly
>> motivated:
>> ex1:cow rdf:type skos:Label;
>>   skos:plainLiteralValue "cow"@en;
>>   dcterms:created "2007-09-09".
>> ex2:cow rdf:type skos:Label;
>>   skos:plainLiteralValue "cow"@en;
>>   dcterms:created "1903-05-05".
>> instead
>> obi:study_trial rdf:type skos:Label;
>>   skos:plainLiteralValue "study"@en;
>>   obi:forCommunity obi:ClinicalCommunity.
>> obi:study_nutri rdf:type skos:Label;
>>   skos:plainLiteralValue "study"@en;
>>   obi:forCommunity obi:NutrigenomicsCommunity.
>> As another example, consider the recording of the association of
>> lexical terms with concepts that would be derived from text mining.
>> In that case we would like to record the fact that the synonym is
>> sanctioned by a particular publication. Thus the skos:Label is the
>> lexical form by which the entity is cited in the paper.
>> Please pass this back to the SWD, and feel free to follow up or have
>> someone else from SWD follow up.
>> Best,
>> Alan
>> -Alan
>>> At 09:08 PM 11/14/2007, you wrote:
>>>> Hi Daniel,
>>>> We had discussed the issue of associating information with synonyms,
>>>> such as provenance, or recording the community that uses the term.
>>>> You mentioned that you would talk to the SWD group about whether this
>>>> as acknowledged as a requirement for SKOS  and if not, whether it
>>>> could be.  I'm wondering whether you got to that, and if so what the
>>>> response was.
>>>> Regards,
>>>> Alan

Received on Monday, 19 November 2007 22:10:03 UTC