Chris Bizer wrote: > Hi all, > > Dan Brickley, Bastian Quilitz (HP Labs Bristol), Alex Polleres (Universidad ^^^^ Axel ;-) > Rey Juan Carlos) and I meet for lunch at ESWC Tuesday and discussed about > database to RDF mapping, SPARQL endpoint descriptions and federated SPARQL > queries. > > Dan Brickley has some connections to the MusicBrainz people and we discussed > mapping their database to an SPARQL-endpoint, which might inspire some nice > mashups. The problem is that D2R server currently answers all queries > (including dump database) which opens up a wide gate for denial-of-service > attacks. We want to fix this in the next version. Anybody good ideas for a > "query capability restriction language"? > > Bastian Quilitz has implemented a version of ARQ which routes queries to > different data sources. His code supports distributed joins. He is also > working on a format for SPARQL-endpoint descriptions and promised to publish > something about his work soon. He knows Richard Cyganiak and they want to > hack a demo that answers SPARQL queries over several relational database > which are mapped to SPARQL endpoints using D2R server. Maybe we can also > reuse some of his work on SPARQL-endpoint descriptions for restricting query > capabilities. Dear all, Just to wrap up my personal interest in this work: We are currently working on an EU project which shall eventually rely on distributed RDF stores for service communication, where this work would fit, and I could imagine fruitful cross-fertilization: http://www.tripcom.org Particularly, I am personally interested in the connection SPARQL & Rules, being a member of the RIF WG. We briefly talked about optimizing distributed query plans and I promised to send some pointers on this. The work which might be relevant/interesting here is: José Luis Ambite, Craig A. Knoblock. Flexible and Scalable Query Planning in Distributed and Heterogeneous Environments. AIPS 1998. These people also gave a full-day tutorial on this topic at ICAPS 2004: http://www-rcf.usc.edu/~skoenig/icaps/icaps04/tutorial3.html > We further discussed requirements for a SPARQL endpoint description language > (there seams to be plenty). This is a very interesting topic, particularly describing e.g. local completeness of a source on particular information etc., see above links, which can be exploited for query plan generation. I plan to investigate whether how the described techniques allow for easy adoption in large scale web environment. best, Axel > As TODOs, we agreed that everybody should add requirements and related work > to the following two wiki pages: > > Database to RDF mapping http://esw.w3.org/topic/RdfAndSql > SPARQL endpoint description > http://esw.w3.org/topic/SparqlEndpointDescription > > Later, we want to aggregate the requirements into a set of evaluation > criteria for comparing the different approaches. > > Dan, Alex and Bastian, did I forget something? > > Cheers > > Chris > -- Dr. Axel Polleres email: axel@polleres.net url: http://www.polleres.net/Received on Tuesday, 20 June 2006 12:23:09 GMT
This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 8 January 2008 14:17:05 GMT