- From: Sigfrid Lundberg, NetLab <siglun@gungner.lub.lu.se>
- Date: Thu, 18 Apr 2002 14:30:45 +0200 (CEST)
- To: Libby Miller <Libby.Miller@bristol.ac.uk>
- cc: Dan Brickley <danbri@w3.org>, "R.V.Guha" <guha@guha.com>, www-rdf-interest <www-rdf-interest@w3.org>
On Thu, 18 Apr 2002, Libby Miller wrote: > > > I'm sorry, but at least currently, the semantic web has a very weak case > > for use in a information retrieval context. > > > > Having just spoken about RDF query at Netlab to the digital libraries > community[1], I'd have to agree with this statement, and as Siggie says, > it's largely a matter of absence of tools, both in terms of > functionality (especially substring matching) and in terms of ease of > use (or lack of). That might be the most important problem, yes. ... > I'd disagree that RDF query has to be very complex to be useful, but the > case for people to use RDF at this moment is not strong, so the issue is > somewhat moot. My argument was not that RDF querying is too complex to be useful, I'm claiming that the human effort required to formulate a query (analysis of the semantic structure, i.e., the structure of the graph) to a triple database is about the same as the effort needed to analyse the syntax some ordinary XML to put bits in pieces into your application. To be more clear -- this is about budget and economy. Assume that there exist 100 metadata DTDs. Then, what is the marginal cost for normalizing a hitherto unknown metadata syntax (metadata syntax number 101) into your 'syntactic' metadata store? Then compare this with the opposite situation. Assume that we have 100 RDF schemas for metadata. Then what is the marginal cost for entering a new semantic structure (a new graph structure) into a 'semantic' (triple) metadata store and extend the triple matching/inferencing to support querying of graph type number 101. Sigge
Received on Thursday, 18 April 2002 08:27:08 UTC