RE: unreification

"Unreification"... what an interesting thought.

Bill dehOra wrote:
> But it doesn't tell me what to do to unreify a statement. .
> So, what can I expect to happen when: 
> 
> a: I remove Rx?
> b: I remove any of the added four triples?
> c: I remove S'
> 
> Can I assume that any removal operation above has the side
> effect of removing other statements (the four added triples)? 

You cannot assume this. The RDF spec does not address retraction of triples.
As for your different cases, I would say that

a: In some way nodes (resources, such as Rx) only exist as far as they are
being referenced by triples; removing a node without removing the triples
containing it would not make sense.

b: In essence you did not really add any information by reifying S', so
removing the added triples does, again, not really change anything (unless
you also made some statement about Rx, and even then only with respect to
some reasoning you do).

c: You can remove S' and keep the other triples. In fact, this is a very
important case we considered when we introduced reification in the RDF
model. Reified form without the existence of the "original" triple allows us
to discuss (describe) statements we do not consider as facts.

> Is it explicitly stated anywhere what the correct
> behavior is? And if it isn't, should the behavior of
> any RDF processor that allows removing statements that
> reify other statements be specifically outlined ? I
> think it's probable that RDF Models which allow
> unreification need some form of transaction semantics
> or truth maintenance. 

Hmm... again, we did not address behavior in any way. So from the spec's
standpoint, there is no such thing as "correct" behavior. From the
standpoint of some imagined consistency of an RDF model, transaction
semantics for reification would be nice. Truth maintenance makes sense if
you are actually reasoning from the facts you have.

Kind regards,

	- Ora

--
Ora Lassila, <ora.lassila@nokia.com>
Research Manager
Agent Technology, Nokia Research Center / Boston
+1 (781) 993-4603 (please note new email & phone number!)

Received on Thursday, 16 November 2000 11:31:28 UTC