Re: Dublin Core and multiple element concerns

> S-B is critical for anybody who wants to write
>	<dc:date>2000-12-23</dc:date>
> and wants tools to flag an inconsistency when they write
>	<dc:date>not the lexical form of any date</dc:date>

Well, if we only had S-B that is all that datatyping will do.
Relating the string "2000-12-23" to the day that was my  brother's 35th
birthday happens (untidily) within the application and not within RDF.

Other ways that the same effect can be achieved include:
- using P
- using TDL global idiom
- using Pat's flash bermuda triangle model theory with all bells and
whistles
- using the syntactic transform that behind the scenes sticks in an extra
rdf:value arc.

All of these have what I perceive as an advantage of actually doing some
model theory concerning dates.

This is why I haven't yet given in on tidyness. I could settle for tidyness
if we are using the syntactic transform, but if, as I think is likely, some
account of datatyping and DC + PRISM + RSS + ... is needed then we will need
one of the above (S-B included).

As I see it the WG has responsibilities both to:
  - the metadata community, who don't do inference, and don't need a model
theory, and don't much care about underlying semantics of a bit of RDF (for
whom not having a model theory for datatyping is likely to be acceptable)
and to
  - the logic community, who have been impressed with the MT so far, and who
I fear would be disappointed at a datatyping solution that is outside the
MT.


Jeremy

Received on Friday, 8 February 2002 09:56:45 UTC