RE: Representing USE-OR in SKOS/XL/ISO25964

Hi Osma,



The iso-thes depicts the compound equivalence relationship as detailed in
red below.

This allows to infer the iso-thes:plusUseTerm and iso-thesUFTerm (USE+ and
UF+ respectively).  See blue arrows.

The iso-thes ontology provided USE+ and UF+ are extensions of
skos-xl:labelRelation.





Whereas the relationships (USE+ and UF+) exists among the terms (see
iso-thes:plusUseTerm and iso-thesUFTerm respectively), no direct relation
is inferred between the concepts ex:A (or ex:B) and the
SplitNonPreferredTerm.

So this does NOT cover the solution 2 and 3 statements making
ex:pitch(“pitch”) an alt label of
ex:A and of ex:B.



Regards,



Johan De Smedt



> -----Original Message-----

> From: Osma Suominen [mailto:osma.suominen@helsinki.fi]

> Sent: Tuesday, April 22, 2014 4:55 PM

> To: public-esw-thes@w3.org

> Subject: Representing USE-OR in SKOS/XL/ISO25964

>

> Hi all!

>

> I have seen many thesauri include a USE-OR relationship, that is, a

> pointer from an ambiguous term (homograph) to several alternative

> meanings (terms or concepts). For example (taken from ISO25961-1,

> section 8.5),

>

> pitch

> USE audio frequency

> OR  gradient

>

> ISO25964-1 recommends against this practice, and instead suggests adding

> qualifiers to avoid having ambiguous terms in the thesaurus. So you

> would only have "pitch (sound)" and "pitch (steepness)" as preferred
terms.

>

> However, some thesauri I'm familiar with (e.g. NAL Thesaurus, CAB

> Thesaurus, YSO) like to include pointers to this kind of either-or

> combinations, for various reasons, including the need to support legacy

> metadata which use the ambiguous term, or to store provenance

> information. How could this be represented using SKOS / SKOS-XL /

> iso-thes / RDF in general? I can think of three ways:

>

>

> 1. Plain SKOS, just use altLabels

>

> ex:A a skos:Concept;

>    skos:prefLabel "pitch (sound)"@en ;

>    skos:altLabel "pitch"@en .

>

> ex:B a skos:Concept;

>    skos:prefLabel "pitch (steepness)"@en ;

>    skos:altLabel "pitch"@en .

>

> This is simple, but the ambigous term "pitch" only appears as a literal,

> so it doesn't get a URI (which could be useful, e.g. when converting

> legacy textual metadata that uses the ambiguous term), and you can't

> easily include provenance information about "pitch", such as the date

> when the split was made.

>

>

> 2. SKOS XL, one shared label instance

>

> ex:A a skos:Concept ;

>    skosxl:prefLabel ex:pitch_sound ;

>    skosxl:altLabel ex:pitch .

>

> ex:A a skos:Concept ;

>    skosxl:prefLabel ex:pitch_steepness ;

>    skosxl:altLabel ex:pitch .

>

> ex:pitch_sound a skosxl:Label ;

>    skosxl:literalForm "pitch (sound)"@en .

>

> ex:pitch_steepness a skosxl:Label ;

>    skosxl:literalForm "pitch (steepness)"@en .

>

> ex:pitch a skosxl:Label ;

>    skosxl:literalForm "pitch"@en .

>

> This way the ambiguous label gets at least a URI, though it's not a

> skos:Concept, which may or may not be a good thing depending on the

> requirements.

>

>

> 3. iso-thes SplitNonPreferredTerm with custom USE-OR property

>

> ISO 25964 has a class SplitNonPreferredTerm which can be used to

> represent "an 'imagined' concept that may exist in a user’s mind but is

> not present in the thesaurus". Though this class is normally used with

> the USE+ relation, one could coin a custom USE-OR property (below called

> ex:orUseTerm) and use it instead, like this:

>

> ex:A a skos:Concept ;

>    skosxl:prefLabel ex:pitch_sound ;

>    skosxl:altLabel ex:pitch .

>

> ex:A a skos:Concept ;

>    skosxl:prefLabel ex:pitch_steepness ;

>    skosxl:altLabel ex:pitch .

>

> ex:pitch_sound a skosxl:Label ;

>    skosxl:literalForm "pitch (sound)"@en .

>

> ex:pitch_steepness a skosxl:Label ;

>    skosxl:literalForm "pitch (steepness)"@en .

>

> ex:pitch a isothes:SplitNonPreferredTerm ;

>    skosxl:literalForm "pitch"@en ;

>    ex:orUseTerm ex:pitch_sound, ex:pitch_steepness .

>

> This is very similar to solution 2, but adds explicit links from the

> ambiguous label to the recommended labels. One could also link directly

> to the concepts, which in my view would make sense, but this is now

> modelled in the same way as USE+ in iso-thes, so the relationship is

> between labels only.

>

>

> What would you do?

>

> -Osma

>

> --

> Osma Suominen

> D.Sc. (Tech), Information Systems Specialist

> National Library of Finland

> P.O. Box 26 (Teollisuuskatu 23)

> 00014 HELSINGIN YLIOPISTO

> Tel. +358 50 3199529

> osma.suominen@helsinki.fi

> http://www.nationallibrary.fi