W3C home > Mailing lists > Public > public-esw-thes@w3.org > August 2008

Re: SKOS Comment (Reference WD June 2008) - broaderTransitive < broader, narrowerTransitive < narrower

From: Antoine Isaac <aisaac@few.vu.nl>
Date: Sun, 03 Aug 2008 14:52:04 +0200
Message-ID: <4895A9F4.2010806@few.vu.nl>
To: public-esw-thes@w3.org
CC: public-swd-wg@w3.org

Hi,

I agree that if we had a thesaurus conforming to thesaurus standards, 
then one of the options 1-2-3 could be valid (of course 1 is a favorite ;-)
But clearly not all the things that are in the scope of SKOS will meet 
one of these conditions.
As said in other mails, SKOS could be used to publish things that are 
not thesauri (even for doing this you have to combine SKOS with 
dedicated extensions). and I think that a lot of the KOSs that advertise 
themseleves as thesauri will not meet the conditions as well. Of course 
you can say that they are not standard thesauri, but that's a different 
matter.

Best,

Antoine


>
> On Sat, 2 Aug 2008 at 14:33:13, Simon Spero <ses@unc.edu> wrote
>> On Sat, Aug 2, 2008 at 9:51 AM, Antoine Isaac <aisaac@few.vu.nl> wrote:
>
>> That's why we de-coupled what is not controversial ("this broader was 
>> asserted in that thesaurus, and we publisher of the thesaurus want 
>> you to know it") from what is useful but should not mess up with 
>> initial assertions (broaderTransitive)
>
>
>> Lets try and clarify things
>
> In the scenario described above,  publisher P asserts:
>
> *A* broader *B*
>
> (Less ambiguously expressed as "A hasBroaderConcept B")
>
> What relationship is being asserted between the extensions of *A* and 
> *B* (*
> A* and *B*) ?
>
> 1)  *A ** B* ?   (All A's are B's )
>
> 2)  |*A* * B*|  > |*A* \ *B*| ? (more A's are B's than are not B's)
>
> 3)  a.(a  *A*  a  *B*)   b.(b  *B*  b  *A*) ?  ( At least one 
> A is
> a B and at least one B is not an A)
>
> [Your logic symbols do not travel well in email - better to just use 
> words]
>
> To conform to thesaurus standards, if the "broader" relationship is 
> generic, then your relationship 1) must apply. The usual example is
>
> Case 1):
> parrots BT birds
>
> valid, because all parrots are birds, while only some birds are parrots.
>
> Cases 2) and 3):
> parrots BT pets
>
> invalid, because only some parrots are pets and some pets are parrots. 
> It seems that these two cases are distinguished only by the degree of 
> overlap of the concepts - they would each be represented by RT, if the 
> relationship is to be shown at all. The RT relationship is somewhat 
> subjective, as "association" is a matter of judgement, following 
> guidelines only.
>
> If the relationships in the published thesaurus do not conform to 
> these standards, it's anybody's guess what the relationships are, or 
> whether they are transitive or not. I don't see how the SKOS version 
> of such a thesaurus can include any additional statements specifying 
> that there are some valid "broaderTransitive" relationships without 
> examining the concepts and determining what valid relationships exist 
> between them - i.e. reworking the relationships in the thesaurus. 
> Reconstructing the thesaurus in this way would be much more than just 
> encoding it in a SKOS format.
>
> Leonard
>
Received on Sunday, 3 August 2008 14:14:43 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:39:00 GMT