Re: Dropping RDF mapping from microdata spec

On 27 Jul 2011, at 09:17, Henry S. Thompson wrote:
> Jeni Tennison writes:
>> Ian is proposing dropping the RDF mapping from the microdata spec (see first comment at):
>> 
>>  https://plus.google.com/u/0/105458233028934590147/posts/3MdCwFNKPfX
>> 
>> Since the presence of that part of the microdata spec is a primary
>> cause of conflict between it and RDFa, and the RDF that is produced
>> by it doesn't match people's expectations (about which Tim filed a
>> bug), I assume that we would support such a change?
> 
> I don't think _dropping_ it is what we want -- we want it _fixed_,
> don't we?


On 27 Jul 2011, at 14:06, Jonathan Rees wrote:
> I think there are many ways to look at this, but one theory I've
> entertained is that microdata bears the same relation to RDF that,
> say, XML or CSV does - it's just another information source that has
> to be reverse engineered and then re-expressed if it wants to be used
> by in an RDF context. Therefore, dictating a single standard way to do
> the re-expression may be as premature and/or inappropriate as it would
> be for CSV or XML. If some RDF is to capture the meaning of some
> microdata, then someone has to come up with a theory that accounts for
> what the particular microdata means, and such a theory might be
> subtle, nontrivial, or context-dependent - at the very least it
> requires some thought.  And even if a standard correspondence were
> desired by the (or an) RDF-using community, that method would probably
> better be specified by those affected, rather than by a working group
> that is obviously not sympathetic to RDF.
[snip]
> So my impulse is to agree with Ian and with you. But I could also
> agree with Henry: If you don't prescribe a re-expression of microdata
> as RDF now, then maybe any hope of doing so later will be doomed.

I suppose that it might be a problem that without a generic mapping defined, different processors end up with different methods of mapping microdata to RDF. But I also agree with Jonathan that the definition of the mapping would be best done by people who care about producing RDF that works well in other RDF serialisations and with the rest of the RDF toolchain.

I know that Michael Hausenblas and others have been working on doing a mapping from schema.org to RDF; perhaps that might work as the basis for a Community Group that could look at defining a mapping.

> (I'm assuming throughout that microdata and RDFa remain separate
> things, and this may change.)
> 
> I didn't think that the microdata to RDF translation is the 'primary
> cause of conflict' in the microdata/RDFa debate; it's just one piece
> of it. Even without any translation at all, we would still have the
> questions of whether there is duplicate functionality, whether that
> matters, and whether anything ought to be done about it.


Sure. I meant *direct* conflict, as in the RDF being generated by a microdata processor based on the mapping in the microdata spec being inconsistent from the RDF generated by a RDFa processor working over the same page. There's still the "should there really be two Recommendations addressing the same use cases in different ways?" question.

Jeni
-- 
Jeni Tennison
http://www.jenitennison.com

Received on Wednesday, 27 July 2011 16:12:40 UTC