W3C home > Mailing lists > Public > www-rdf-logic@w3.org > December 2001

Re: DAML+RDFS: potentials for simplifications?

From: Pat Hayes <phayes@ai.uwf.edu>
Date: Tue, 11 Dec 2001 12:25:07 -0600
Message-Id: <p05101001b83bfd4d512d@[65.212.118.193]>
To: Jeen Broekstra <jbroeks@cs.vu.nl>
Cc: www-rdf-logic@w3.org
>On Thu, 29 Nov 2001 Joachim.Peer@unisg.ch wrote:
>
>>  You say, a result of the simplified syntax is, that a
>>  processor would need a "set of conventions". Yes, but how
>>  does this differ from the current situation?
>
>It doesn't, and I think we are touching upon the core here:
>by using a set of conventions (RDF!) that seems likely to be
>shared by a broader community we are increasing
>interoperability. Possibly at the cost of not having the
>most optimal model, but this is a typical result of a
>compromise. You can't have your cake and eat it, too.
>
>>  Are you aware of any DAML interpreter which has no
>>  hardcoded set of DAML specific instructions?
>
>Yes. Any RDF parser for example. Or the RDF Schema query
>engine that we are currently building. Any RDF or
>RDFS-specific tooling in fact.

I think this answer is disingenuous. Any RDF parser can parse 
DAML+OIL , but it parses it as RDF, not as DAML+OIL. In order to 
'interpret' (which I take to mean, be able to draw valid conclusions 
from) the DAML+OIL, one needs to know more than just RDFS: one needs 
to know how the DAML+OIL syntax is encoded into RDF. For example, one 
needs to know that some of the RDF is asserted as part of a DAML 
assertion, but other pieces of RDF are assertions about the syntax of 
the DAML assertions.

>The OIL - RDFS relationship has been set up in such a way
>that compatibility works in both directions: any OIL spec is
>valid RDF Schema, and an RDF Schema is a valid OIL ontology.

In the very narrow sense of 'valid' which means 'can be processed 
without being rejected as syntactically incorrect'. Even in this 
sense, there are well-formed pieces of RDF that are not legal 
encodings of any DAML+OIL assertion.

>We have tried to layer OIL on top in such a way that as much
>knowledge as possible is captured within the original RDF
>Schema model, thus allowing "maximum" knowledge sharing with
>less semantically aware (read: RDFS-only) tooling.

Maybe that is true of OIL, but it is very arguable for DAML+OIL.

>That is
>the suspected added bonus, which I personally think will
>prove very valuable in a heterogeneous environment, where
>lots of different levels of expressiveness (from simple RDF
>to DAML+OIL) are required for different tasks.
>
>Btw, there is a paper which deals with this issue, which
>might interest you. The paper is from some time ago and
>deals with the original OIL rather than DAML+OIL, but the
>arguments still hold I think:
>
>     http://www.cs.vu.nl/~jbroeks/papers/www10.pdf
>
>>  take for example the definition of the unionOf-construct of DAML:
>>
>>  <rdf:Property rdf:ID="unionOf">
>>    <rdfs:label>unionOf</rdfs:label>
>>    <rdfs:comment>
>>      for unionOf(X, Y) read: X is the union of the classes in the list Y;
>>      i.e. if something is in any of the classes in Y, it's in X, and vice
>>  versa.
>>      cf OIL OR
>>    </rdfs:comment>
>>    <rdfs:domain rdf:resource="#Class"/>
>>    <rdfs:range rdf:resource="#List"/>
>>  </rdf:Property>
>>
>>  this definition tells an RDF aware application only half
>>  of the story... It tells nothing about the semantic
>>  implications of this construct. So if one wants to have a
>>  DAML aware agent, one will need to tell it explicitly what
>>  e.g "unionOf" means.
>
>True. But the RDFS-aware application will still know that
>there is a relation called UnionOf.

Right, but that is false in DAML+OIL. There is no such relation: that 
'relation' is part of the syntax. See Peter Patel-Schneider's recent 
postings to the joint committee and the subsequent discussions:
http://www.daml.org/listarchive/joint-committee/0934.html

>It will know that which
>classes exist, it will know their instances and it will
>understand the subsumption relation.

On the contrary, it will have the wrong idea about which classes 
exist (compared to the idea that is encoded in the DAML+OIL 
assertions) and moreover there is no way to tell it which ones do 
exist, since RDFS is not currently capable of making the required 
distinctions.

Pat Hayes
-- 
---------------------------------------------------------------------
IHMC					(850)434 8903   home
40 South Alcaniz St.			(850)202 4416   office
Pensacola,  FL 32501			(850)202 4440   fax
phayes@ai.uwf.edu 
http://www.coginst.uwf.edu/~phayes
Received on Tuesday, 11 December 2001 13:25:18 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:52:41 GMT