W3C home > Mailing lists > Public > www-rdf-interest@w3.org > November 1999

Re: Cambridge Communique

From: Sankar Virdhagriswaran <sv@crystaliz.com>
Date: Fri, 12 Nov 1999 20:19:15 -0500
Message-Id: <199911130120.UAA27663@hunchuen.crystaliz.com>
To: "Ralph R. Swick" <swick@w3.org>
CC: www-rdf-interest@w3.org
Ralph,

>XML Schema cares more about the form of the expression, RDF Schema
>(and RDF Model&Syntax) care more about the meaning of the expression.
>

This is true. However, representing knowledge does not come easy. This is
made harder by the syntax chosen by RDF. Additionally, the cognitive
overhead of having to learn two different languages establishes a conceptual
barrier where there is no reason for the conceptual barrier to exist except
for historical reasons. Therefore, if one wants to represent semantics and
yet is interested in leveraging the adoption of XML-Schema by DBMS and other
repository vendors, one will represent that semantics with limited support
provided by XML-Schema.

As an example, one only has to look at the DBMS community. For the longest
time now, it has been known in the research community that representing
semantics (knowledge) is very important to mediating heterogeneous DBMSs.
Yet, the approaches that have taken off there have been based on structural
approaches. IMHO, this is because the knowledge representation languages
(such as KIF) that were proposed to address the 'semantics' problem setup a
conceptual chasm that was never crossed by most developers. For good or bad,
RDF has established such a chasm (syntax, dense documentation, poor use
cases, no compelling community effort that shows off the use in simple
cases, etc., etc.)    
 
>ought to make that happen.  But multiple object classes that provide
>different views on a single data structure is also a bit of technology
>that we all understand how to work with in our toolset.
>

Could you please point to an example which has the property of getting
widely adopted by novice users. One only has to look at CORBA's language
mappings to see the kind of conceptual barriers to entry for novice users.
Aren't we talking about 'markup' languages that have the property that they
have to be understood and adopted by a large community that is not formally
trained in first order predicate calculus or 'reification' ;-).

IMHO, there is a big gap that RDF tried to fill, but lost sight of along the
way: a *simple* markup language for meta-data. That gap still exists.
However, time has passed us by. Yes, it is simpler than XML-Schema + XLink
for this purpose, but not by much. Also, the lack of smooth progression from
RDF to XML-Schema or XML-Schema to RDF is *now* a problem for adoption of
RDF. 

This is one of the reasons I believe you are not getting any feedback on
request for APIs, etc. One has only so much mind space. The cognitive
overhead in learning about things that seem to do the same things in
different ways in the context of other standards (DOM, XLink, XPath, to name
a few) at a certain point makes a pragmatic consumer to toss out the glory
of 'semantic web'. 

If you believe my analysis is correct, then there are only two courses of
action left. Fall in line with other standards that have well identified and
pragmatic purposes so that there is not a cognitive overhead for the
consumer. Or, assemble a simple and small sub-set with the associated
*revolutionary* community to get it going. 

Knowledge representation is a tough problem. Let us not make it harder.

Sankar

 
Received on Friday, 12 November 1999 20:18:39 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:51:42 GMT