W3C home > Mailing lists > Public > public-esw-thes@w3.org > February 2017

Re: Using iso-thes to publish 1:n-relations between skos:Concepts from different concept schemes

From: Antoine Isaac <aisaac@few.vu.nl>
Date: Fri, 17 Feb 2017 01:02:23 +0100
Message-ID: <58A63D8F.1090705@few.vu.nl>
To: "Svensson, Lars" <L.Svensson@dnb.de>
CC: "public-esw-thes@w3.org" <public-esw-thes@w3.org>
Hi Lars,

Yes the XKOS approach has some cons. It's a case where handling the provenance of correspondences than had a more important priority than
easing the consumption of simple data. In fact the XKOS pattern is similar to the ones used in the Ontology Alignment domain in the Semantic Web community.

I guess the decision on using MADS/RDF also depends on how the 'groupings' of concepts can be seen as 'real' SKOS concepts rather than ad-hoc, application-specific combination. In a way, this is a bit a case of pre-coordination vs post-coordination. In the MACS case MADS is a rather good fit as it's about headings which are largely designed for being combined.

Best,

Antoine

On 16/02/17 19:40, Svensson, Lars wrote:
> Hi Antoine,
>
> On Thursday, February 16, 2017 1:37 AM, Antoine Isaac [mailto:aisaac@few.vu.nl] wrote:
>
>> Sorry for the delay answering this email.
>
> No problem: I'm looking for a good solution, not a quick one...
>
>> You are right in your understanding of ISO-THES' CompoundEquivalence is rather
>> between terms/labels and concepts.
>>
>> MADS/RDF may have something better, with madsrdf:ComplexType and
>> madsrdf:componentList
>> http://www.loc.gov/standards/mads/rdf/
>>
>> But perhaps the closest thing available is the XKOS pattern for correspondences:
>> http://www.ddialliance.org/Specification/XKOS/1.0/OWL/xkos.html#correspondences
>>
>> (in fact I've delayed this mail because I wanted to review XKOS)
>>
>> I'm not sure it does all you need, though. XKOS doesn't have 'typed correspondences'
>> of the form of 'OR' and 'AND' combinations, which were identified as a requirement in
>> the SKOS context.
>> The names they use are also not so great. See
>> https://github.com/linked-statistics/xkos/issues/31
>
> Yes, I've been looking at XKOS, too, particularly for publishing the MACS dataset. I do like the approach since it makes the relation between the two concepts a first class citizen, so that you don't need to use reification if you want to add metadata to it. OTOH it makes it harder to use in a linked data environment when you publish one vocabulary and simply wants to link to another one (e. g. GND to LCSH or GND to STW). So currently my tendency would be to use mads/rdf.
>
> Thanks for your comment. If others have insights, please let me know.
>
> Thanks,
>
> Lars
>
>> On 25/01/17 16:47, Svensson, Lars wrote:
>>> Dear SKOS-Community,
>>>
>>> Here in the DNB we're currently revisiting how we publish our thesaurus mappings in
>> RDF with a focus on how to publish 1:n-relations (i. e. where a concept in one KOS is
>> mapped to two or more concepts in another KOS). We don't publish those relations yet
>> since we haven't found a best practice. I've been looking at madsrdf which sort-of-
>> works and last week I delved into iso-thes which has CompoundEquivalence which looks
>> like a good starting point. However, if I understand the documentation correctly
>> CompoundEquivalence can only be used between _terms_ (within one KOS?) and not
>> between _skos:Concepts_.
>>>
>>> I'm aware that this is an old discussion [1] and probably not resolved yet. However,
>> any insight you can provide would be most helpful!
>>>
>>> [1] https://lists.w3.org/Archives/Public/public-esw-thes/2011Jun/0007.html and
>> subsequent messages...
>>>
>>> Best,
>>>
>>> Lars
>>>
>>> *** Lesen. Hören. Wissen. Deutsche Nationalbibliothek ***
>>>
>
Received on Friday, 17 February 2017 00:02:59 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 19:46:52 UTC