W3C home > Mailing lists > Public > public-esw-thes@w3.org > February 2007

Re: FW: [SKOS] semantics

From: Mark van Assem <mark@cs.vu.nl>
Date: Mon, 12 Feb 2007 17:21:08 +0100
Message-ID: <45D093F4.8090404@cs.vu.nl>
To: public-esw-thes@w3.org

Hi Alistair,

Nice document! I appreciate the logic rigour, but wouldnt it be more 
readable for most audiences to use a rule format or "constraints"?
(Depends of course on the goals of the document)

But I do have some questions:

- Semantic Condition 1 (preferred lexical label cardinality)

I am starting to wonder about this one. From a technical perspective 
it is no longer necessary because we can create two URIs for homonyms.

Furthermore, WordNet does not have one preferred term for a synset. 
They all have equal status, so because of this SKOS requirement WN 
cannot be mapped correctly without violating this constraint.

This choice in WN makes sense: words are just ambiguous and different 
words can point to the same thing. What differs is the frequency and 
context with which two alternative terms are used.

So is this constraint really necessary? Isn't the pref/alt distinction 
more related to display as Jacob suggests?

- A Note on the Transitivity of skos:broader

  "other applications such as retrieval systems that perform 
graph-based reasoning (where the number of steps in a path is an 
important metric) will not want to treat skos:broader as transitive."

Aren't you assuming then that there are NO transitive triples in the 
hierarchy already? Either this should be a Semantic Condition or 
broader should be transitive.

Cheers,
Mark.

> Semantic Condition 4 (broader/narrower inverse)
> Semantic Condition 5 (related symmetric)
> Semantic Condition 6 (broader/related disjoint)
> 
> Ok. By the way about non-symmetric ("see also") links between concepts?
> There should be a recommended way how to encode them.
> 
> 
> SKOS Semantic Condition 7: broader circularity
> 
> This should be an additional condition at a higher level then the other
> condition. Of course in most cases circles do not make sense but they
> may occur (the category system of Wikipedia is an example - there
> regularly circles occur). Circles are less easy to reduce so they should
> be not recommended but not forbidded by default. Just do it the same way
> as the Transitivity of skos:broader condition
> 
> 
> SKOS axiomatic triples
> 
> skos:prefLabel rdfs:range rdfs:Literal.
> skos:altLabel rdfs:range rdfs:Literal.
> skos:hiddenLabel rdfs:range rdfs:Literal.
> 
> I don't agree on this. How about prefered sign language labels, prefered
> image representations? Annotated labels that are resources of their own?
> 
> 
> Greetings,
> Jakob
> 

-- 
  Mark F.J. van Assem - Vrije Universiteit Amsterdam
        markREMOVE@cs.vu.nl - http://www.cs.vu.nl/~mark
Received on Monday, 12 February 2007 16:21:34 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:38:55 GMT