Hi Dominique,
Welcome in this list! I agree with Nina, this is a perfect use case for the semantic web which offers to view the web as a distributed database. Then RDF and SPARQL can be good friends. However, there will be performance drawbacks for the web join to happen. There are developers of the popular Sesame RDF database in this list that might be able to provide additional information: http://www.aduna-software.com/technology/sesame Another popular open source RDF store is Mulgara: http://www.mulgara.org/ <http://www.mulgara.org/index.html> Im just not sure if/how distributed SPARQL queries effectively work, but if you end-up having all data in memory on the client-side to be able to do the join then it isnt worth it and you might have to solve your use case in a different way. This reminds me of the limitations of Google App Engine storage which doesnt support joins for performance/scalability reasons This definitely forces you to rethink your query strategy :) Best regards, Jerome Louvel -- Restlet ~ Founder and Technical Lead ~ <http://www.restlet.org/> http://www.restlet.org Noelios Technologies ~ <http://www.noelios.com/> http://www.noelios.com De : Nina Jeliazkova [mailto:[email protected]] Envoyé : lundi 1 mars 2010 07:51 À : [email protected] Objet : Re: rest practice : how do you "join" linked resources from 2 providers ? doj wrote: Hi, first hello to Restlet team, i know some of you guys. My question is for you and everybody else implied in practicing REST and remoting in general. A first provider catalogs information on a domain being a reference for others; for instance, all information on the partners involved in your business. The second provider handles data referencing partners, let's simply take the parameters for the partners for a given business process. I'd like to implement full-text search for those parameters: find parameter values based on either example parameters values or example partner labels. With tables in SQL world, I can "join" the tables and specify criterias on both (perhaps it's not a good practice by the way). How do you handle this with remoting and resource oriented services ? For this kind of use case, "algorithmic join" of XML/other representations retrieved on the fly can seem unnecesseraly complicated, so people often replicate reference data, in order to typically to make this stuff with "locally stored" index information. What do you think of a "on the fly" scenario by the way ? Or perhaps this full-text multi-source search idea is not a good one ? If resources expose RDF representations and SPARQL endpoints, one can perform SPARQL queries remotely , even embed different service providers within the same SPARQL query. Not sure if the SERVICE syntax is a SPARQL standard, but IMHO is a close solution for what is being asked. (Example from http://jena.sourceforge.net/ARQ/service.html) PREFIX : <http://example/> <http://example/> PREFIX dc: <http://purl.org/dc/elements/1.1/> <http://purl.org/dc/elements/1.1/> SELECT ?a FROM <mybooks.rdf> { ?b dc:title ?title . SERVICE <http://sparql.org/books> <http://sparql.org/books> { ?s dc:title ?title . ?s dc:creator ?a } } Best regards, Nina Jeliazkova thanks for your point of view. Dominique. ------------------------------------------------------ http://restlet.tigris.org/ds/viewMessage.do?dsForumId=4447&dsMessageId=2456493
