- From: Gabe Beged-Dov <begeddov@jfinity.com>
- Date: Sun, 26 Nov 2000 10:50:44 -0800
- To: Seth Russell <seth@robustai.net>
- CC: RDF-Logic <www-rdf-logic@w3.org>
Seth Russell wrote: > I can see three possible solutions: > > 1) The sets discussed in [The RDF Model] are really > multisets and triples are not unique. Therefore the triple > I see at robust.ai is not the same triple I see at the > Electoral College; the one can be true and the other false. > A triple is unique only within a semantic island and we need > to be able to express in RDF to which island it belongs. > > 2) We go down the slippery slope where every agent must view > triples as reified statements. When it reads a triple it > sees a reification quad - when it writes a triple it writes > a reification quad. > > 3) We toss the law of the Excluded Middle. 1 and 2 can be seen as two sides of the same capability, i.e. to track the triple/statement occurence in terms of the source document. As I pointed out ([1]), the M&S supports (requires?) the use of reification to track the semantic islands (aka documents) that statements were uttered in. Manipulating the raw triples/statements is initially the path of least resistance for implementations but I think it is actually a much slippier slope that actually figuring out how to handle reification correctly and efficiently. What current implementations are calling the "model" is just the set of statements that occurred in a particular source document. We should be talking about the model which is the set of Bags of reified statements that occurred in the source document (semantic island). If we did the latter then the problem of tracking statement occurrences in a model that contained multiple source documents would not be an issue since the bags and the reified statements would all be labeled with URIref whose URI would be the source document. The fact that this is what the M&S says we should be doing (at least according to my divination) makes it even more critical that we thoroughly consider this approach to solving some of the scaling issues related to manipulation of statements occuring in multiple "semantic islands". I'm sorry if I'm hammering on this, but I'd really like some of the brain power of the list participants to be applied to the question of whether we are actually doing what the M&S says we should or allowing percieved implementation costs and intertia of existing implementations drive our thinking. Cordially from Corvallis, Gabe [1] http://lists.w3.org/Archives/Public/www-rdf-interest/2000Nov/0366.html -- --------------------------- http://www.jfinity.com/gabe
Received on Sunday, 26 November 2000 12:50:57 UTC