W3C home > Mailing lists > Public > semantic-web@w3.org > December 2010

RIF/SPARQL Semantics (Was -- Rule usage description...)

From: Adrian Walker <adriandwalker@gmail.com>
Date: Sun, 12 Dec 2010 09:23:11 -0500
Message-ID: <AANLkTinb3bSJaLvvvoe-bQtye4JUffDUAL-4KSpePq=S@mail.gmail.com>
To: Semantic Web <semantic-web@w3.org>
Hi Ivan & All --

There would seem to be an opportunity here to move the W3C approach to rules
and SPARQL onto a firmer semantic base.

The basic idea is to define what answers a rule system or SPARQL should give
to *any *question, based on *any *set of triples, using a logical model
theory.  The theory then works as a "gold standard" for implementers.  W3C
could provide test suites.

Looking backwards for a moment, SQL suffers from  the lack of such a base or
standard.  There is a query which produces different answers over the same
data in two of the leading  SQL implementations (and both answers are
intuitively wrong).  That's not fatal for SQL, since one can 'program around
it', but a similar flaw is extremely serious for RIF/SPARQL, since it's
supposed to work the same everywhere on the Web.  How would we ever trust
the answers?

An early version a model theoretic standard is in [1,2].   There's an
illustration of how it can work in practice with RDF in [3,4].

Hope this helps.

                                     -- Adrian

[1]  Towards a Theory of Declarative Knowledge, (with K. Apt and H. Blair).
In: Foundations of Deductive Databases and Logic Programming, J. Minker
(Ed.), Morgan Kaufman 1988.
http://oai.cwi.nl/oai/asset/10404/10404A.pdf

[2]  Backchain Iteration: Towards a Practical Inference Method that is
Simple   Enough to be Proved Terminating, Sound and Complete. Journal of
Automated Reasoning, 11:1-22

[3]  www.w3.org/2004/12/rules-ws/paper/19

[4]  www.reengineeringllc.com/demo_agents/RDFQueryLangComparison1.agent
Received on Sunday, 12 December 2010 14:23:47 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 21:45:40 GMT