- From: Pat Hayes <phayes@ihmc.us>
- Date: Fri, 11 Aug 2006 11:44:33 -0700
- To: Jorge Pérez <jperez@ing.puc.cl>
- Cc: franconi@inf.unibz.it, public-rdf-dawg-comments@w3.org
>Pat Hayes wrote: > >[...] > >> >> [Later: And the other example in >> http://lists.w3.org/Archives/Public/public-rdf-dawg-comments/2006Aug/0005 >> makes the basic point more effectively.] >> >> However, consider the following example: same query, but against the graph >> >> {(a p a) (X p X) (X p b) } >> >> Here, I suggest, the answers ?q->a, ?q->X would >> be correct, in spite of the apparent redundancy, >> because the third triple clearly distinguishes >> (what is known about) X from (what is known >> about) a; so the redundancy is indeed only >> apparent. > >I totally agree with you that the answer ?q->a, ?q->X is correct in this >case, but I think that your example is not related to the core of this >discussion. I agree, but I wanted to make this second point in order to head off some possible alternative suggestions for how to fix the bug which you identified. >The "bug" in BGP E-matching is originated by redundancy in the >dataset side (and queries about such redundancies) and has nothing to do >with redundancy in the answer side, that (as your example shows) one not >always want to eliminate. Quite. This point has been contentious in the past, hence my example. I am glad we are in agreement on the correct behavior in these cases. >Your example is "a good example" for the definitions, the dataset has no >redundancy and the BGP E-matching definition (with simple entailment) >gives the same solutions as the subgraph matching approach, so the "bug" >is not present. Indeed, I think that it is a theorem that in the case of >lean datasets the two approaches are always equivalent. I also think this, but I no longer have total trust in my intuitions :-) . Pat -- --------------------------------------------------------------------- IHMC (850)434 8903 or (650)494 3973 home 40 South Alcaniz St. (850)202 4416 office Pensacola (850)202 4440 fax FL 32502 (850)291 0667 cell phayesAT-SIGNihmc.us http://www.ihmc.us/users/phayes
Received on Friday, 11 August 2006 18:44:45 UTC