W3C home > Mailing lists > Public > public-esw-thes@w3.org > January 2011

RE: How to distinguish unique and non-unique prefLabels?

From: Ford, Kevin <kefo@loc.gov>
Date: Wed, 12 Jan 2011 11:05:46 -0500
To: Jakob Voss <jakob.voss@gbv.de>, "public-esw-thes@w3.org" <public-esw-thes@w3.org>
Message-ID: <1D525027B29706438707F336D75A279F152C4FD00A@LCXCLMB03.LCDS.LOC.GOV>
Hi Jacob,

Short of creating a new property, Joachim's suggestion seems good [1].

The solution you proposed - defining skos:uniquePrefLabel rdfs:subPropertyOf skos:prefLabel - might result in an integrity violation after inferencing.  Based on the example you provided, and please correct me if I'm wrong, you could end up with two same-language skos:prefLabels for the same resource after inferencing.  This would violate the skos:prefLabel integrity condition that "a resource has no more than one value of skos:prefLabel per language tag."[1]

The only other thing I could think to do would be to add a "no linguistic content" language designation ("zxx") to the second, "unique", to-be-used-with-the-KOS-as-thesaurus skos:prefLabel property (I'm proposing this, but can't say that I'm sold on the idea myself).  This would permit the two labels to peaceably coexist as skos:prefLabel properties.  At best, this keeps the information in the skos namespace, but still fails to lend any meaningful distinction to your "unique" prefLabel.  You'd still have to define a system rule that says "when using the KOS as a thesaurus, look for the 'zxx' prefLabel," but this solution would unambiguously identify a concept within a KOS and its applications.



[1] http://lists.w3.org/Archives/Public/public-esw-thes/2011Jan/0016.html
[2] http://www.w3.org/TR/skos-reference/#L1567

From: public-esw-thes-request@w3.org [public-esw-thes-request@w3.org] On Behalf Of Jakob Voss [jakob.voss@gbv.de]
Sent: Wednesday, January 12, 2011 08:47
To: public-esw-thes@w3.org
Subject: How to distinguish unique and non-unique prefLabels?


Mapping a classification (not a thesaurus) to SKOS, I stumbled upon the
problem of distinguishing unique and non-unique preferred labels. The
concepts in our classification have

1. a unique notation (unique)

2. a main label (not unique, as it's no thesaurus)

3. an extendend main label (unique by adding a qualifier)

3. possibly alternative labels

Here is a simple example with the non-existing property
"skos:uniquePrefLabel" for the extended, unique label:

   skos:notation "15.80" ;
   skos:prefLabel "history"@en ;
   skos:narrower <15.87> .

   skos:notation "15.87" ;
   skos:prefLabel "USA"@en ;
   skos:uniquePrefLabel "History (USA)"@en ;
   skos:altLabel "United States"@en .

I am searching for a best practice to encode both types of preferred
labels: the non-unique but short label, best used in combination with a
notation and/or in a hierarchical view if the KOS is used as
classification, and the artificial but unique label, best used if the
KOS is used as thesaurus.

The best solution that I found so far, is creating

skos:nonUniquePrefLabel rdfs:subPropertyOf skos:prefLabel ;
   skos:scopeNote "a preferred label, that is not unique for some
concept scheme" .

skos:uniquePrefLabel    rdfs:subPropertyOf skos:prefLabel ;
   skos:scopeNote "a preferred label, that is unique for some concept
scheme" .

but I hesitate to create just another RDF property that nobody else
uses. Any suggestions?


P.S: The same can be applied to notations, but non-unique notations are
probably less common than non-unique preferred labels.

Jakob Vo▀ <jakob.voss@gbv.de>, skype: nichtich
Verbundzentrale des GBV (VZG) / Common Library Network
Platz der Goettinger Sieben 1, 37073 G÷ttingen, Germany
+49 (0)551 39-10242, http://www.gbv.de
Received on Wednesday, 12 January 2011 16:07:50 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 19:46:06 UTC