- From: Kingsley Idehen <kidehen@openlinksw.com>
- Date: Thu, 28 Jan 2010 08:11:13 -0500
- To: Richard Light <richard@light.demon.co.uk>
- CC: Christoph LANGE <ch.lange@jacobs-university.de>, Linked Data community <public-lod@w3.org>, Vyacheslav Zholudev <v.zholudev@jacobs-university.de>, Florian Rabe <f.rabe@jacobs-university.de>
Richard Light wrote: > In message <4B605247.6030201@openlinksw.com>, Kingsley Idehen > <kidehen@openlinksw.com> writes >>> And that mixing and matching Linked Data, Topic Maps and full-text >>> XML (etc.) can potentially take us to some exciting new places. >>> (But then that's hardly news. I remember discussing the value of >>> URNs for this sort of purpose some time around 1994.) >> I agree, and I am simply saying: Topics are Things too, and they can >> exist in EAV/CR model relationship graphs. > > Sure. Has anyone actually tried that: storing assertions in a triple > store, but treating them as a "virtual Topic Map"? It's an approach > which might yield benefits - and there are possibly paradigms other > than Topic Maps which could usefully be virtualized on top of > triple-store data. > >> I think we are exhibiting semantic pedantry here based on our >> individual world views, which is fine, since this is really what this >> whole gig is supposed to facilitate :-) > > Not just semantic pedantry, Kingsley, fun though that is ;-) My > inclusion, above, of full-text XML adds, I think, another dimension to > this debate. I think all dimensions are expressable via relations using existing infrastructure be it the relationship collection associate with <link rel="{some-relation}" .../> or claims in RDF triples. I believe you are describing an "association". > > In all of this clever flattening of heterogeneous data onto the level > playing field which is RDF, we are enabling information retrieval: the > answering of questions. And that's fine. The information that is > flowing around and available to us in this context is all RDF: simple > assertions. They are really EAV based assertions, RDF is a framework that uses this model to facilitate a webby variant of EAV via incorporation of URIs (generally) and Generic HTTP URIs specifically re. Linked Data. The actual graph model of EAV has been with us for a very long time. > However, the user who originally asked the question might not be > satisfied with this type of information: they might want a story, not > facts. Where does the story come from? The document that contains it. Said document has a URI (like any other Item of interest). > > If each subject of discourse (whose URI we already know) can be > delivered as full-text XML via the standard 303 redirection mechanism, > that gives us one way of grabbing the full story in > machine-processible form, and including whatever parts of it are > relevant in our responses to questions To the degree I understand what you are trying to articulate, I do believe this can happen as is, even via SPARQL with the XML processed from a literal object in a triple. Kingsley > > Best wishes, > > Richard -- Regards, Kingsley Idehen President & CEO OpenLink Software Web: http://www.openlinksw.com Weblog: http://www.openlinksw.com/blog/~kidehen Twitter: kidehen
Received on Thursday, 28 January 2010 13:11:50 UTC