W3C home > Mailing lists > Public > www-rdf-comments@w3.org > April to June 2003

Re: FORTH COMMENTS ON RDF Schema: Last Call

From: Vassilis Christophides <christop@ics.forth.gr>
Date: Fri, 4 Apr 2003 09:13:10 +0300 (EEST)
Message-Id: <200304040613.JAA14377@athena.ics.forth.gr>
To: phayes@ai.uwf.edu
Cc: danbri@w3.org, brian_mcbride@hp.com, www-rdf-comments@w3.org

>>Dan
>>
>>Thanks for the answer. Yes it is quite quite intresting to open the
>>discussion on the two issues.
>>
>>My understanding is that only if you have in mind the whole picture of
>>the RDF/S specifications you may arrive to useful design choices.
>
>That is likely to be true, but surely that is to be expected of most 
>specifications.

This is exactly the difference between a standardization and a research
 activity.

>>Furthermore, up to know most of the choices are based on the open
>>world assumption. However, I din't see any useful reasoning service
>>defined on top.
>
>There are several RDF(S) and OWL reasoners available. Details of some 
>of them can be found from the relevant W3C websites, or by tracing 
>links from the DAML or OIL  websites.

Could you please indicate me what are the reasoners for DAML+OIL:
besides Fact I also know Triple (http://www.daml.org/tools/). What are
the useful reasoning services which are supported: subsumption
checking (C1 < C2), instance checking (a in C)? It is the core of the
functionality required by real scale SW applications? In addition,
what are the performances and scalability of these tools? main-memory,
single user, no log complexity access to data. 

For these reasons I believe that the XML technology is much more
mature than the RDF(S)/DAML+OIL and the industry already has already
been convicted that it worths to put money in the so-called syntactic
web. 

>>On the contrary I have seen a lost of useful services
>>(validation, efficient storage, query optimization) based on the
>>"closed world assuption".
>
>I am not sure how the closed world assumption would provide for 
>efficient storage, but you may be right that validation could depend 
>on it. 

Having a reasonable type system for any data model is the main support
to ensure consistency of data (aka validation), efficient storage (aka
secondary memory support) and query evaluation (aka
optimization). These are the main lessons of more than 30 years of
research in DB.

>However, one should not assume that RDF is *incompatible* with 
>a closed world assumption. RDF reasoning is not itself based on a 
>closed world assumption because there is no way to 'close' the world 
>of a web reasoner, in general. But if a particular application or 
>community knows that it can rely on a closed world assumption, and 
>has some way to know what the boundaries of that closed world are 
>(such as the information in certain set of databases, say) then one 
>could make RDF reasoners which rely on that assumption or shared 
>knowledge. They would make inferences which are not strictly 
>RDF-valid, but that is likely to be common. OWL reasoners, for 
>example, make RDF inferences which are not strictly RDF-valid. A 
>closed world assumption could thus be considered to be a semantic 
>extension to RDF. Open-world reasoning is valid in a closed world; 
>the open world assumption is basically just another term for 
>logically valid reasoning, and closed-world assumptions represent 
>extra assumptions in addition to general logical validity.

Can we simply support safe negation on both the values and the
structure of RDF descriptions? 

>>It is quite disappointing that after 4 years of passionate discussions
>>on the open world assuption the only real systems that someone can use
>>and implement bits of the Semantic Web are based on the "closed world
>>assuption". This is quite scary!!!
>
>It might be scary if it were true, but it is not true.
>

Honestly, besides syntax translators I din't see any profound
technological achievements the last 4 years. I would like real to
suggest to RDF/DAML+OIL communities to take a more close look to the
research and development efforts conducted the same period for XML. 
Then one can really understand that the situation is really scary
regarding the future of the Semantic Web activity.


>Pat Hayes

Vassilis
Received on Friday, 4 April 2003 01:17:52 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Friday, 21 September 2012 14:16:32 GMT