Re: peer-to-peer was Re: Distributed querying on the semantic web

From: Graham Klyne <GK@ninebynine.org>
Subject: Re: peer-to-peer was Re: Distributed querying on the semantic web
Date: Tue, 27 Apr 2004 11:31:25 +0100

> [Transferring a discussion from RDF-IG...]
> 
> At 16:16 24/04/04 -0400, Peter F. Patel-Schneider wrote:
> >I also believe that there already is sufficient machinery in the Semantic
> >Web to support a combination of (2) and (3), namely owl:imports.  Yes, I
> >would like something better, perhaps to allow for publishers of information
> >to provide sub-document groupings of information.  Bijan Parsia and I have
> >a poster paper at WWW2004 on this topic, available at
> >http://www-db.research.bell-labs.com/user/pfps/publications/meaning.pdf
> >(unless you read this message soon after it is posted, in which case the
> >slow web publishing mechanisms I have may not have got around to noticing
> >it).
> 
> Reading this paper, I find the approach presented to be compelling.  The 
> clear explanation of potentially complex issues is helpful.  I like the 
> articulation that dissent is allowed and consent is easily expressed.

Thanks.

> I note that the work [1] by Jeremy Carroll, et al, might well provide a 
> means for referencing parts of web documents (though that does presume that 
> the source document is somehow partitioned to allow the desired granularity 
> of selective inclusion).

Agreed. This is a possible way to allow sub-document inclusion. 

> It's not clear to me the extent that you are proposing that meaning can be 
> imported from sources opaque to Semantic Web reasoning.  

Hmm.  Actually I am proposing that *no* Semantic Web meaning can be
imported from sources opaque to Semantic Web reasoning.

> You allow for 
> possible augmentation to "Semantic Web meaning to be contingent on a set of 
> Semantic Web languages that the system understands" -- by "Semantic Web 
> meaning", I understand you to mean the aspects of meaning that manipulated 
> by a Semantic Web reasoner.  

Yes.

> But in talking about communities of consent, 
> there will always (I believe) be aspects of meaning that are not amenable 
> to such processing;  it seems to be reasonable that a SWeb document should 
> be able to indicates its author's intent to be conformant to such meanings, 
> maybe by reference to a human-readable description of such meaning.  

Sure, but in my view this intent is only part of the Semantic Web to the
extent that it is accessible to Semantic Web reasoning processes.

> Maybe 
> this is what you mean by "There is nothing ... that prevents software 
> systems from augmenting, or even replacing, the Semantic Web meaning with 
> their own notions of meaning"?

I mean here that applications will generally go outside the (core) Semantic
Web meaning in various ways.  


> Do you see in this proposal the basis of a mechanism to determine whether a 
> document coded in RDF is intended to include the additional language 
> constraints on interpretation that are required of an OWL document?  

Well, I think that MIME type may turn out to be a useful first step in this
process, except that OWL has the same MIME type as RDF.

> I 
> could imagine a URI whose importation would be equivalent to a declaration 
> of adherence to OWL semantics, not just RDF semantics, but for which there 
> is no machine-readable document that actually asserts the additional 
> constraints.

I don't think that this is the way to go.  Instead I would prefer some
meta-data indication (such as MIME type).

> I would guess that if, as you suggest, the imports notion is picked up by 
> RDF, then it would import the RDF meaning of a document rather than its OWL 
> meaning.  

Well, an RDF processor would only get the RDF meaning.  An OWL processor,
reading an RDF file that imported another file that had OWL constructs,
should, I think, pick up the OWL meaning of these constructs.

> Another approach might be to augment the import relation to 
> permit expression of the semantics being imported.  Following a pattern of 
> "interpretation properties" [2] suggests maybe an open-ended set of such 
> properties (maybe all sub-properties of "rdfs:seeAlso" [3]), each of which 
> operates on a defined level of semantics;  e.e. rdf:imports relates RDF 
> graphs; owl:imports relates OWL ontologies, etc.

Hmm.  This would work, but I think that it would be better to just point at
an information source (e.g., http://www.foo.ex/doc) and use some other
mechanism (content negotiation?) to get representations of the information
(e.g., one with RDF MIME type vs one with OWL MIME type).

> #g
> --
> 
> [1] http://www.wiwiss.fu-berlin.de/suhl/bizer/SWTSGuide/carroll-iswc2004.pdf
> 
> [2] http://www.w3.org/DesignIssues/InterpretationProperties.html
> 
> [3] http://www.w3.org/TR/rdf-schema/#ch_seealso
> 
> 
> ------------
> Graham Klyne
> For email:
> http://www.ninebynine.org/#Contact

peter

Received on Wednesday, 28 April 2004 10:42:19 UTC