Tod worte
> I guess this is possible if we don't introduce multiple sets of words
> (which I know we will want to do) at the same time.
Exactly what do you mean with "multiple sets of words". If they are
seperate "words" then, for sure, they should not belong to the same
schema.
> The version would allow folks that don't require the latest and greatest
> the ability to keep functioning without harm.
Then you have to build into the architecture how to manage legacy. Models
collected over time, using different namespaces for exactly the same
semantics. So, not only will you have to use inferencing power for
crosswalking IEEE LOM, GILS and DCMI, you will have to use it for
cross-walking DCMES versions.
> If we had only one name for DCT (say http://purl.org/dc/types/) then we
> could not distinguish between contents of the sets (significant adjustments,
> additions, or refining) that WGs so passionately argue. Maybe we can reach
> an equilibrium (the ideal set of words that make everyone happy), and we can
> all go home...:-) IMHO we should plan for the corrections and/or work left
> on top of whatever we publish.
Actually, if the Types WG finally delivers a proper type ontology, we have
all the RDF(s) machinery for building that on top of the old DCT1
vocabulary. I don't think that is a problem. The same is true for elements
and qualifiers. The price we have to pay for some schema handicraft in
this context is low, in comparison with the price we will have for
maintenance of RDF legacy metadata over time.
Sigge
|