[RDFTM] comments on survey

Some partial comments (i.e. I've not finished reading) on
http://www.w3.org/2001/sw/BestPractices/RDFTM/survey-2005-02-24

I've read David's comments and largely agree with him.  I'm hoping
to understand some of his comments better during f2f discussion.
Others I won't repeat here but may amplify during the discussion.

First, thanks Steve, Fabio, and the rest of the RDFTM Task Force for
a helpful first document.

It is important to consider RDF and OWL together when the
combination addresses some of the translation issues.  Otherwise
a reader may be misled into thinking that some aspect has not been
treated by either Topic Maps or SemWeb community when in fact
it has.  For example, in 2.2.1 under 'identity' we find the statement
"RDF only allows a single identifier".  In the specific contexts
OWL sameAs and inverseFunctionalProperty can permit multiple
identifiers.  Under "Association roles" the parenthetical remark
makes the implication that OWL is separate whereas I think the
full utility of any translation will certainly use OWL.

The N-ary associations paragraph should cite the WG n-ary relations
Working Draft.

Editorially, I caution against making commitments for future deliverables
in this document.  Language such as "expects to" is better than "will"
when referring to documents that are not yet published.

In the 2.3.2 RDF2TM test case it might be useful to add foaf:homePage
in order to expose the features of Topic Maps subject identifiers (at
least, I expect that would do so.)

In both 2.3.1 and 2.3.2 it would be useful to denote which portions
come from the instance data and which would normally be expected
to come from a shared document, such as the RDF namespace
document (for the rdfs:label statements in 2.3.2.)

In section 3 there are several instances of the phrase "it is not clear".
It is not clear to me ( ;) ) which of these statements are paraphrasing
the associated author(s) of the specific proposals -- i.e. the original
author identified some lack of specificity, whether the statements
are being made by the authors of this survey, or whether the statements
are being made on behalf of the Working Group (in which case the WG
might be expected to engage in some more 'deep thought' to see if
clarity might emerge.)

In 3.2 much of the difficulty of the approach chosen follows from the
choice of [PMTM4].  Has there been a better model for Topic Maps
written since that work was published?  I wouldn't expect the TF to
reproduce the work of Lacher and Decker with a different (and
hopefully more complete) expression of the Topic Maps data model
but this survey should cite such a model if it does exist.

more later, perhaps ...

-Ralph

Received on Thursday, 3 March 2005 14:07:34 UTC