[topicmapmail] XTM-RDF conversion

[Dan Corwin]

Dicheva, Darina wrote:

> We are currently implementing an XTM-RDF conversion for TM4L.
...
> I would much appreciate any feedback on translating TM association into
> RDF statements. 

I got into a bioinformatics group a couple of years back, one deeply
committed to OWL-DL and other W3C standards.  I often suggested they
might learn useful things from XTM, and as part of that, I once tried
to formally redefine "association" as a design pattern in RDF/OWL-DL
based on blank nodes.  Here (in TURTLE) is the pattern and one usage:

   http://www.lexikos.com/exhibit/cms/biopax/assoc-topic.ttl

Minor complications do arise in the pattern, and scoped occurrences
or names seem hard, but for associations, scope just adds an extra
triple (as does type), while the binary-relation limitations of RDF
evaporate.  More discussion and examples are here and nearby:

   http://biopaxwiki.org/cgi-bin/moin.cgi/AssociationPattern

My colleagues fed such models into Pellet and were able to "infer"
successfully by using them.  But I am sorry to say they've shown
no huge upsurge of interest (yet), possibly as blank nodes often
get weak RDF-tools support.  Hope TM4L can help to change that.

cheers,
Dan Corwin