- From: Jeremy Carroll <jjc@hpl.hp.com>
- Date: Tue, 03 Jan 2006 10:41:16 +0000
- To: Misha Wolf <Misha.Wolf@reuters.com>
- CC: semantic-web@w3.org, www-tag@w3.org, newsml-2@yahoogroups.com
Misha Wolf wrote: > The section seems to offer two semi-viable options: > > - Use rdf:ID to identify a statement and then make statements about > the identified statement. > > - Use some other, application-specific, way to identify statements > but then don't expect anyone else to understand it. > I liked Charles' reply concerning EARL, which took the second approach. This avoided the semantic problems with RDF reification, and like any RDF application specific vocabulary could be understood using standard RDF approach: a) the minimal meaning is given using the RDF Semantics b) to understand (informally) the intent of any of the terms, click on the URI and (hopefully) arrive at a definition (usually with an rdfs:comment that gives human readable text) I feel that the interoperability loss that you seem to fear is a chimera. When you say "but then don't expect anyone else to understand it" ... The NewsML vocab itself is only intelligible to people who have bothered to try and understand it. The SW gives a method for allowing them to understand it, by clicking on the terms. The SW also gives a minimal formal semantics to avoid some basic misunderstandings. If RDF reification worked then what it would offer would be a general purpose mechanism for making statements about other statements, noting that these other statements have a time and place and a propositional attitude and not just a logical meaning. RDF reification does not achieve that; and so the current situation is that there is no usable, standard, general purpose mechanism for making statements about statements. (aside: Personally I think the work done on named graphs by myself, Bizer, Stickler and Hayes, addresses this general purpose need; but introduces new mechanisms for which there is no standard support.) So, as whenever you want a module in your code/schema that does something general purpose, if there isn't one available that meets your needs you have to write your own. It is generally cheaper to write one that addresses your needs narrowly scoped, but good to plan it in such a way that it could be extended/modified to address more general needs. One approach would be to have a class x:Assertion that is perhaps defined as a superclass of earl:Assertion. You could then factor out of EARL the bits that you wanted, e.g. assertedBy, but not the test case specific stuff. This would be a step towards a more general purpose mechanism, reusing a bit of EARL, but primarily focusing on the NewsML problem and not the general one. my 2c Jeremy
Received on Tuesday, 3 January 2006 12:37:20 UTC