Re: [ALL] PORT documents for internal review - 1/2

> The following documents are submitted to the working group for internal
> review:
> 
> (A) SKOS Core Vocabulary Specification (2004-12-17 version)
> http://www.w3.org/2004/02/skos/core/spec/2004-12-17.html
> 
> (B) SKOS Core Guide (2004-11-25 version)
> http://www.w3.org/2004/02/skos/core/guide/2004-11-25.html
> 
> (C) Quick Guide to Publishing a Thesaurus on the Semantic Web (2004-11-17
> version)
> http://www.w3.org/2004/03/thes-tf/primer/2004-11-17.html
> 
> The nominated reviewers for these documents are:
> 
> Mark van Assem (on behalf of Guus Schreiber)
> Tom Baker
> 
> Reviews should be posted to this list by 10 January 2005.

My review focuses on Document B -- the 40-page overview of SKOS
Core -- though my comments have implications for the other two.

Overall, this is excellent, careful work.  I want to say
this up-front because, after a close reading of the document,
I end up raising quite a few points of detail.

My second posting will raise points of wording and
presentation.  This posting covers three more fundamental
issues:

1. Reaching the intended audience

   As discussed in the telecon of 16 December [1]:
   >   The guide is human-readable intro - how to use it:
   >   features of vocabulary, with examples.  In the last
   >   telecon, we agreed to make it accessible to non-RDF
   >   people, but proved to be nearly impossible to write -
   >   would have been extremely long.  Rather, we restrict
   >   the scope to people who basically understand RDF,
   >   then if we want to present porting issues, we will
   >   do that in a separate doc which explains basic
   >   concepts (not yet written).  From there, we can
   >   look at developing add'nl method notes.

   A separate document on "basic concepts" will be a useful
   thing, but in the meantime a bit more introduction is
   perhaps needed in the SKOS Core Guide itself.

   The Guide does assume that the reader is RDF-literate.
   However, it presents that RDF in the form of RDF/XML
   serialization syntax.  While the Introduction emphasizes
   that SKOS Core is not "an XML syntax for concept schemes",
   this is done to make the point that N3/Turtle or N-Triple
   could be used just as well -- and not to reinforce the
   more basic point that "what is fundamental to RDF is the
   graph model" [RDF-PRIMER].

   One or two simple node-arc diagrams right at the beginning
   of the draft might be a simple and readable way to present
   the "basic concepts" behind SKOS.

   For example, the example concept from the Quick Guide
   ("Economic cooperation") illustrates in itself some basic
   features of SKOS Core: skos:Concepts, related to other
   broader or narrower skos:Concepts, with preferred versus
   alternate labels.  Presenting this one example as a simple
   diagram with labeled arcs and nodes could be a good way
   to present the basic idea.

   The introductory message, then, could convey something
   like the following message:

      Thesauri represent semantic relations among concepts
      [insert "Economic cooperation" example here, along with
      citations for BS8723, ISO 2788, and other thesaurus
      standards].

      Here is how the example looks as an RDF graph using the
      SKOS Core vocabulary [inser a node-and-arc diagram here].

      If your vocabulary has a similar structure, you will
      be interested in reading this Guide because it will
      tell you how you can express your vocabulary in, or
      translate your vocabulary into, an RDF model using the
      SKOS Core vocabulary.  Using the RDF model will allow
      your vocabulary to be linked to or merged with other
      data structures by RDF applications.

2. What SKOS Core "is"

   The Abstract begins:

        "SKOS Core is a supporting RDF Vocabulary..."

   To me, this choice of wording raises several questions
   that are not really answered in the rest of the text.

   Someone familiar with RDF -- the target audience of
   the draft -- might correctly take an "RDF Vocabulary"
   to be something like "a vocabulary of terms usable as
   Properties and Classes in the RDF model".  In the absence
   of a definition, however, the reader could confuse it with
   "The RDF Vocabulary" ("a set of URI references in the rdf:
   namespace" [2]).  Some readers, concluding that SKOS Core is
   only relevant to people who are already "using RDF", might
   stop reading right here.  A definition of "RDF vocabulary"
   up-front, with a pointer to [3], could address this.

   But is the SKOS Core Guide really primarily about a
   vocabulary?  Or is it really about a particular data
   model based, in turn, on the RDF model?  Reducing SKOS
   Core to the vocabulary alone seems a bit like reducing
   RDF to "The RDF Vocabulary".  Saying that SKOS Core is a
   "supporting" vocabulary makes one ask: supporting what?

   Rather, describing SKOS Core as a "model" for expressing
   knowledge organization structures such as thesauri could
   perhaps correct this narrow perspective, shifting the
   reader's attention to the model of entities being described
   ("skos:Concepts" and relationships between them) and how
   the vocabulary "supports" that model.

3. Ownership and maintenance of SKOS

   In the Vocabulary Management task force, we are trying to
   formulate (and illustrate) a best-practice guideline to
   the effect that vocabulary maintainers should "articulate
   and publish maintenance policies for the Terms and their
   URI references".  It is not clear from the documents (in
   particular the SKOS Core Vocabulary Specification) who
   is ultimately taking responsibility for the maintenance
   of the SKOS vocabulary.  Is W3C implicitly assuming that
   responsibility?  I'm wondering to what extent the SWBPD
   working group needs to address these questions as a basis
   for any recommendations it may want to issue.

   As a related issue, the Vocabulary Spec is generated from
   the RDF representation, implying that the RDF representation
   is canonical and the Web document is derived.  Yet it is
   the Web document that we are reviewing, presumably to
   assign the Web document some sort of status in the W3C
   context.  Which representation is primarily the object of
   maintenance?  This relationship between the Web document
   and the underlying RDF representation should perhaps be
   addressed in the Introduction.

[1] http://lists.w3.org/Archives/Public/public-swbp-wg/2004Dec/0099.html:
[2] http://www.w3.org/TR/rdf-mt/
[3] http://www.w3.org/TR/rdf-schema/

-- 
Dr. Thomas Baker                        Thomas.Baker@izb.fraunhofer.de
Institutszentrum Schloss Birlinghoven         mobile +49-160-9664-2129
Fraunhofer-Gesellschaft                          work +49-30-8109-9027
53754 Sankt Augustin, Germany                    fax +49-2241-144-2352
Personal email: thbaker79@alumni.amherst.edu

Received on Thursday, 23 December 2004 09:47:08 UTC