W3C home > Mailing lists > Public > public-sw-meaning@w3.org > October 2003

Re: Provenance/Justification

From: Jim Hendler <hendler@cs.umd.edu>
Date: Wed, 8 Oct 2003 19:01:41 -0400
Message-Id: <p05200f18bbaa46508562@[]>
To: pat hayes <phayes@ihmc.us>, Sandro Hawke <sandro@w3.org>
Cc: public-sw-meaning@w3.org

At 4:38 PM -0500 10/8/03, pat hayes wrote:
>>   > >Yes.  The triples on the bio page are thrown in with the rest.  (I
>>>   >suggested we may want to tag where they came from for
>>>   >explanation/trust reasoning, some day soon.)
>>>   Yes, soon. I think tags like this would be handy for many engines, in
>>>   fact.  Our graphic RDF/OWL editor needs to keep them around for more
>>>   mundane reasons: even for writing simple markup, the user often wants
>>>   to be able to check where a concept 'comes from'.
>>>   Maybe that would be a good architectural principle: SW concept names
>>>   should always 'come from' a unique resource, which we might call the
>>>   provenance of the name.
>>Two different but related concepts here.  I was talking about tagging
>>triples; you're talking about tagging names.   I'm thinking about
>>"who/what made this claim".
>OK, sorry I was sloppy there. We need both, actually. A lot of the 
>heat in this discussion seems to come from the cases where they 
>diverge, so A is asserting a triple using B's URI.
>>>   Exactly what this means, and what
>>>   relationships there might be between the resource they come from, the
>>>   representations extracted from that resource using other Web
>>>   protocols, and what the concept name should be understood to denote,
>>>   can all be discussed at more length; but at least that would give us
>>>   a piton to nail into the rock.
>>Actually, I can't separate what you're talking about here from the
>>general "what does a URI mean" question, sorry.
>Oh, please.  Did I use the M-word anywhere in that? Ive been trying 
>to find ways of completely avoiding ever saying 'meaning' when 
>referring to URIs.  Provenances could be genuinely architectural. 
>Meanings are another matter altogether.

Interestingly, the formalism for SHOE (the first web ontology 
language I was involved with) was based on what we called a claims 
logic -- we essentially said that the URIs used in SHOE (and it was 
URI based) were meant to represent claims -- that is

if document D said the SHOE equivalent of
   A rdfs:subclass B
then the knowledge base was updated with something like
  Claims(D, A rdfs:subclass B)

this was routinely rejected whenever we submitted it to a conference 
because the AI folks said it was silly to use a higher order logic. 
However, it had a great feature -- the claims logic remains 
consistent even when the content doesn't (and is also monotonic) -- 
that is, there is no inconsistency in
  Claims(D, X)
  Claims(D, -X)
the inconsistency would be in
which could not happen in SHOE (or on the web, unless one invents a 
mechanism for "not claiming" something - such as using a closed world 
assumption or the like)

just a historical note.
Professor James Hendler				  hendler@cs.umd.edu
Director, Semantic Web and Agent Technologies	  301-405-2696
Maryland Information and Network Dynamics Lab.	  301-405-6707 (Fax)
Univ of Maryland, College Park, MD 20742	  *** 240-277-3388 (Cell)
http://www.cs.umd.edu/users/hendler      *** NOTE CHANGED CELL NUMBER ***
Received on Wednesday, 8 October 2003 19:01:49 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 19:56:01 UTC