- From: David Allsopp <dallsopp@signal.dera.gov.uk>
- Date: Tue, 19 Dec 2000 15:52:53 +0000
- To: "McBride, Brian" <bwm@hplb.hpl.hp.com>
- CC: www-rdf-interest@w3.org
"McBride, Brian" wrote: > > One of the things which struck me lately during one of > our lengthier discussions was the idea that many of our > disagreements may be based on different assumptions > about the applications we have in mind RDF being used > for. > > It might be helpful therefore, to have a list of > canonical use cases so we can say, "well, if you > are trying to do X, then we want ..., but if you > are trying to do Y, then ..." > > So what are you're favourite use cases for RDF? <snip> I'm interested in using RDF for communications and data storage in heterogeneous agent-based systems. The focus here is on agents that were not designed together as a system, but are brought together in an ad hoc way and need to be made to interoperate quickly, in some cases by wrapping a legacy system as an agent. RDF (plus a query language) offers interesting possibilities of using groups of agents in flexible and unforeseen ways . I think this requires a query language that can specify sections of an RDF model to be returned as RDF documents, rather than queries which just return (arrays of) value(s). I might want to know about a particular resource, and may not want to send dozens of query messages (perhaps via a very slow transport mechanism) to extract each attribute of the resource - ideally I want to send one message and get back a serialized RDF model that can be stored or merged, and queried locally. Most current work on querying RDF seems to be oriented towards clients querying database servers (or webpages) rather than a peer-to-peer architecture with multiple RDF databases. In addition, I need to be able to make queries that involve reified statements (handling statements that carry attribution, timestamps, accuracy estimates etc), and obtain such statements from another agent, preserving that reification (perhaps multiple levels) so the origin and nature of the statements is preserved. From what I've heard, most current query proposals don't handle this, at least not in an implementation-independent way, and none allow you to extract a statement plus the triples encoding arbitrary levels of reification: i.e. queries about the triple S = ['John Smith' age ?x] could, if desired, return not only actual triples matching this triple, but quads reifying this statement, together with any additional data "Jane says S", "As of 21st January 1999, S", etc. Regards, David. -- Dr David Allsopp DERA Malvern
Received on Tuesday, 19 December 2000 10:54:20 UTC