- From: Bill de hÓra <dehora@acm.org>
- Date: Mon, 20 Nov 2000 21:27:07 -0000
- To: "RDF-IG" <www-rdf-interest@w3.org>
- Cc: "Dan Brickley" <danbri@w3.org>, "Graham Klyne" <GK@dial.pipex.com>
: Perhaps we need to consider extending the RDF Model. My position is that : the model is unambiguous, but agree that it is perhaps too restrictive for : certain use cases. Again and again we run into problems related to a : monolithic model on a distributed web. If we introduce the concept of a : "space" to provide context to a set of statements, there would be no problem : with multiple statements having the same (p,s,o) as long as they "exist" : within distinct spaces. : : One way to formalize this is to state that: : : There exists a set of spaces (sp)* contained by the global space g. A space : sp contains a set of statements (p,s,o)*. Spaces, predicates and subjects : are resources. An object may be either a resource or a literal string. We're getting somewhere here :). If an sp contains statements and is a set, it is potentially a member of the power set of S, the set of all statements mentioned in the RDFm recc. So you're not saying anything in conflict with RDFm. Top hat. I smell W3c note. If this is the case, it it still ok for these spaces to have URIs? Spaces not having URIs would suck. Because if spaces can have URIs then we can: -implement join/discovery protocols for spaces. -export rdf encoded service descriptions into spaces (for any computational service) -encode requests for services as rdf queries. -have associative memory on the web (at last). -implement danbri's description services. -have a basic architecture for web of trust. -implement Gelertner tuple spaces using w3c standards. We've been discussing this in work for some time now. This is great. This is what I want for xmas. -Bill de hÓra
Received on Monday, 20 November 2000 16:29:36 UTC