Re: MiniSKOS update: back to "Topic" (but thanks for trying out the ConceptCode idea)

Hi everyone,

It was not my preferred one, but I very much prefer Topic over ConceptCode!
Coded classifications are an important case, but there are others code-less concept schemes, and calling their labels 'codes' is quite hard to do [1].

If you need more examples of vocabularies which could be feeding the Topic class:
http://www.w3.org/2001/sw/wiki/SKOS/Datasets
http://datahub.io/dataset?q=format-skos

A quick SKOS mapping question on the current proposal: is codeValue equivalent to skos:notation? It seems to me that it is the case. It would be much useful to state an explicit mapping, especially if schema.org is going to include its own terminology (again!) here.

And finally a general broader issue: having an anchor point for typing concepts in schema.org is certainly good. But soon or later people will ask how to describe these concepts/topics in schema.org, no?
Unless you are intending to make a first step of a more distributed approach to schema.org maintenance and use, where users would be encourage to re-use the properties and classes from other namespaces?
Is it the meaning of
[
  SKOSProperties This class includes properties whose definitions are based on the work of SKOS, see http://www.w3.org/2004/02/skos/ for details.
] [2]
?
There was a lot of discussion earlier about the issue. And http://schema.org/AlignmentObject and earlier problems with Action demonstrate how confusing it is when a specific domain seizes a very generic label for its specific purposes [3].
But I thought that schema.org's approach was rather to include everything needed in one vocabulary...


Best,

Antoine

[1] In fact KOS frameworks have generally moved from 'terms' and 'codes' as their main focus to 'concepts' and 'classes'. Reverting to a low-level name as 'code' would be really counter-productive here, if one wants to convey the "thing vs. string" message. Actually some the mails, e.g.
http://lists.w3.org/Archives/Public/public-vocabs/2013Nov/0183.html
really show the possible confusion (in there only the code matters, as a simple key to use in a search system).

[2] https://dvcs.w3.org/hg/webschema/file/default/schema.org/ext/miniskos.html

[2] In the Linked Data community, 'alignment' means something really different from the LRMI sense. Many things can be aligned--starting with concepts/topics from different enumerations/conceptschemes.

Received on Sunday, 24 November 2013 21:48:26 UTC