Hi Neo4j/TinkerPop people, I am wondering if Neo4j will become more efficient in its representation of RDF named graphs now that it has relationship property indexing. Efficient in terms of space. If I remember correctly, a quad statement is ?2? neo4j relationships. The reason being, there was no way to get the 'g' in 'spog' in an efficient (index) manner without such a representation. Thus, without this ?2?-statement construction, a getStatements(null,null,null,tg:someGraph) would require a scan of all relationships.
Now with being able to index relationship properties, its possible to have tail_node [value: uri/bnode] relationship: [graph: uri/bnode] [label: uri] head_node [value: uri/bnode/literal] - See https://github.com/tinkerpop/gremlin/wiki/Modeling-RDF-as-a-Property-Graph For the Blueprints scene, I'm wondering if it would be good to implement Sail in Blueprints (we currently have Blueprints implementing Sail). As such, we could turn any multi-index graph database into an RDF quad store without requiring the multi-index graph database to even support, know about, or contribute a Sail implementation... (of course, they would have to have a Blueprints implementation). Thoughts? Marko. http://markorodriguez.com _______________________________________________ Neo4j mailing list User@lists.neo4j.org https://lists.neo4j.org/mailman/listinfo/user