- From: Nicolas Chauvat <Nicolas.Chauvat@logilab.fr>
- Date: Sun, 12 Nov 2000 14:54:58 +0100 (CET)
- To: Jason Diamond <jason@injektilo.org>
- Cc: www-rdf-interest <www-rdf-interest@w3.org>, 4suite@dollar.fourthought.com
***PLEASE BEWARE with follow-ups, this is a cross-post*** > I haven't quite been able to wrap my head around how one would use an RDF > model like the DMOZ dumps without loading it into memory or importing it > into a database that you could query against. The question I have is this: > Can an API be devised that abstracts away whether or not the model is loaded > into memory or persisted in a database and still be useful to us as > developers? Ok, this is not exactly RDF, which probably needs a different API than DOM's, but http://www.4Suite.com/ (part of Python's DOM et al. implementation) is having a go at abtracting a DOM document whatever its storage may be (memory, DB or flat file). The API they use is called 4ODS and documentation is available at http://fourthought.com/4Suite/4Ods/. ODS is an ODMG standard for persistent objects. > I remember reading (but can't recall where) that the RDF model is actually > extremely close to the relational model. The author pointed out that columns > are like properties and field like objects where the primary key in each row > was the subject. This obviously doesn't take into consideration repeated > properties and a number of other issues but it did open my mind up to start > thinking about storing a RDF in a more resource-centric manner. One of my > goals for RDF.NET is to explore this approach and see where it ends up. As for storing/querying big xml repositories, there is code to look at and implementation ideas to borrow from at http://www.dbxml.com/, http://sourceforge.net/projects/xdbm/ and http://www.xmldb.org/. -- Nicolas Chauvat http://www.logilab.com - "Mais oł est donc Ornicar ?" - LOGILAB, Paris (France)
Received on Sunday, 12 November 2000 08:59:10 UTC