[humanmarkup] Re: [humanmarkup-comment] [RDFCore] Items (was:HumanMarkup: Paved With Good Intentions)

[Kurt Cagle <kurt@kurtcagle.net>]

>> Relational associations can also be done with topic maps.

>Absolutely, but there are less XTM people in here than RDF people...
Granted, and I don't think we want to go there now anyway. Both Len and Paul
Prueitt (whom I'm kind of sort of channelling for the group) seem to have
problems with the direction of XTM, and I have reservations myself. It's an
option, but not something that should be a major focus for a while.

>XSD is what I wanted to skip for the most part, using RDFS. In RDF,
>there is no need to define a syntactical layer; you just concentrate on
>the actual model you want to define.

I'm kind of ambivalent about this, myself. RDFs doesn't really have a clear
cut notion of metric, and is also something of a black box for most people,
even those familiar with XML syntax. I see data-types (metrics) coming into
this discussion frequently, and I don't think that is going to go away. I
would more likely say that we may be better off parallel developing XSD and
RDF structures with cross-talk between the two, using XSD where it works
best -- as a mechanism for defining metrics -- and using RDF where it works
best -- as a mechanism for defining relationships.

> >In other words, I see RDF as essentially defining two
>> resources (schemas)
> >and a "verb" that ties the two resources together; the XSLT
>> (or DOM, though
>> >I think XSLT is better in this regard) then defines the
>> actions of the
>> verb.
>Sounds interesting but I don't get it right now (I'm at lost at work).
>Could you please provide an example?

At the risk of Sean's wrath (sorry, Sean) the happy example that I gave
earlier in the discussion actually provides one such mechanism. I need to
think about this somewhat; I have a few examples in mind, but it occurs to
me that it may be better to make them fairly rigorous and incorporate them
into a formal paper for the HumanML group.

Blessed Be,
Kurt