- From: Alan Wu <alan.wu@oracle.com>
- Date: Fri, 13 Jun 2008 10:07:25 -0400
- To: Bijan Parsia <bparsia@cs.man.ac.uk>
- CC: OWL Working Group WG <public-owl-wg@w3.org>
Bijan, > >> Bijan, >> >> Sorry for the delayed response. >> >> Seems that we don't quite agree on the how much additional cost by >> leaving axiom triples out. >> I am glad to see at least we agree that it requires a more >> sophisticated implementation. :) Can I ask the WG then to simply >> the mapping so that unsophisticated developers like me >> have an easier time implementing OWL2 in a commercial product. I >> believe that is a very reasonable >> request. > > It needs to be balanced by other considerations. That is fair. BTW, I forgot to mention that adding the axiom triple won't cause a huge expansion of the ontology. Do we truly worry about, say 20%, size increase? > As I've pointed out, it's not clear at all to me that in the situation > you've outlined (lots of annotated triples in a large kb) that you can > *avoid* the need for a sophisticated implementation. If people are > querying for annotations, you have to do something to cope with > mapping the reified triples to the non-reified one. Better to do that > at load time. Well, it really depends. If an implementation chooses to optimize the performance for query/inference over non-reifiied data and put a much lower priority on query over reified data, then such a sophisticated implementation may not be necessary. > Plus, there's a clear bit of advice for people to optimize loading: > Don't randomize your triples. > That is a good advice for tools that generate N-Triples :) Cheers, Zhe
Received on Friday, 13 June 2008 14:09:20 UTC