RE: [humanmarkup-comment] RE: HumanMarkup: Paved With Good Intent ions

["Bullard, Claude L (Len)" <clbullar@ingr.com>]

Compared to HumanML, efforts such as UBL are simple. 
Work on early versions of what will be UBL has been 
going on for over a decade now.   VRML was built 
on the design of Open Inventor.  X3D is built on 
VRML.  HTML is built on GenCoding and then 
SGML.  XML is SGML As Practiced.  RDF 
and the Semantic Web are built on the research 
and experience of DARPA, Stanford University, CMU, 
and people such as Tom Gruber. There are 
no new inventions there nor are any needed. 
Expert systems, ontologies, rules-based AI, 
these are all well understood technologies 
by the experts in the AI field.

Where one has a technology to harvest, 
it is much easier to show "invention" in less 
than a year.  Where one does not, much basic 
research and experimentation is required. If 
the pace seems slow, it is normal.   Schedules 
should reflect that and so far, none have 
been impediments.   
 
We can focus our efforts around the technology 
for which we do have expertise:  schemas.  This 
will have a cohesive effect in reducing 
the surface area of technologies we need to master 
to focus our work on useful products.  If we need 
RDF(s), we can come back to it later after we 
solidify the schema work and have working 
examples that will enable us to clarify how to use 
HumanML with other XML application languages such as 
SVG, X3D and XHTML rendering languages.   

Kurt Cagle made a proposal yesterday 
for the use of xlink attributes.  I asked if 
we should retain and use the abstract definitions 
as a means of deriving the complete set of 
HumanML types.  The technical issue 
we should address for getting our schemas into 
a more usable form is the architecture of the 
schema itself.  

len