[humanmarkup-comment] [RDFCore] Items (was: HumanMarkup: Paved WithGood Intentions)

[Manos Batsis <m.batsis@bsnet.gr>]

Hi all, get a cup of coffee. I'm addicted myself.

Parenthesis. Although I haven't been active due to personal
difficulties, I never said I'm leaving HumanMarkup. Actually, the
incident forced me to wake up and start doing some work. End
parenthesis.

Seems the resignation incident produced excellent results as side
effects. I'm sure that this focus and insight will be produced in a
direct way from now on (well, most of the times). So, thanks to all. I
will skip all things I could have wrote since they don't matter too much
(you all know more or less how I would have responded to everyone in
private level or in this list) and just focus to the points that I find
worthy of your time. 

More formal documents than this will follow and be published, but input
has to be asked for in this list. I will be producing two documents,
developed around a possible but well defined project. One will be the
framework (which I will call Platform from now on) outline in an
acceptable depth of detail. The other one will be the requirements
document. Both documents will be constructed by interaction between
them, taking place in the "[RDFCore] * " thread.. The first draft will
be published in thirty days from now, no matter it's progress. I will be
responsible for the document as I have accepted the role of "technical
architect for RDF" (wow, a hat). My intention includes the formation of
a core team responsible of working over this, without of course cutting
off the rest of the group from the process. We just need the
responsibility thing.
The discussion starts with this post. Note: This thread welcomes non TC
member input ;-)



Platform discussion: RDFCore.
Datatypes/explicit values (meaning literals) etc. Models.
=============================

The reason I strongly suggest we should try to avoid literals (aka
explicit values like "4" or "high") of any datatype in our core Platform
(I do not wish to ban literal values in implementations) in general, is
that it will be extremely difficult for implementors (and of course, us)
to design systems that will be able to react to such values. Even if
they do, all this work will be totally useless as it will represent
custom, non reusable development. What we need of the Platform is it's
ability to grow. Examples of this feature in different implementations
are OOP languages, everything XML (especially RDF related, namespaces
and schemata) etc.
On top of that, resources according to the RDF model are a simple model,
already implemented for us, that introduces benefits which are not all
new; for a small fraction of what I mean, consider the differences of a
system that processes a number of different literal values (of any
datatype) versus resources (as values) of object oriented hierarchies.
Of course, this efficiency can be expanded well beyond the hierarchical
model.

TODO: Investigate possible complications of breaking the hierarchical
model. 
I consider this to be of top priority. This will possibly get off this
list as well.

TODO: Document methodologies of using the resource model for building
applications that intelligently take advantage of this attribute value
model. This may span a third document to publish.



One list, different root threads.
=============================

It has been pointed out that different objectives need their own room.
Since we don't really need a number of lists at this stage, I suggest we
insist on a form of subject threads, at least for the RDF related ones.
For example, to reply to the technical section above please do use
[RDFCore] in the start of your subject i.e.: Subject: "[RDFCore] I want
datatypes...".
It will help whoever is building summaries or just keeping track of the
posts.

Kindest regards,

Manos