Hello,
The Blazegraph team just announced the release of Blazegraph 2.0.0
https://blog.blazegraph.com/?p=977
Blazegraph2 supports both TinkerPop3 (and TinkerPop2). They leverage the work
of Olaf Hartig who wrote a article a few years back about how to map property
graphs to RDF graphs [http://arxiv.org/abs/1409.3288]. This isomorphism allows
Blazegraph (which is an RDF store) to work with property graphs. Moreover, I
was talking with the Blazegraph guys the other day and they were saying how
they would like to leverage TraversalStrategies more so that they can compile
more of their Gremlin traversals into SPARQL queries (as SPARQL is what their
system is optimized for). A "back of the napkin" approach was articulated:
SparqlQueryStep<S,E>
SparqlTraversalStrategy
[GraphStep, MatchStep(…), OrderStep, RangeStep, PropertyMapStep]
[BlazeGraphStep, SparqlQueryStep(match/order/limit), PropertyMapStep]
In general, for all RDF graph providers that leverage TinkerPop3, it would be
great if we had SparqlQueryStep and SparqlTraversalStrategy. This way, they can
come to TinkerPop, implement the low-level "Blueprints" API and then be stoked
that Gremlin traversals (w/ SparqlTraversalStrategy) will compile to use SPARQL
for their ultimate database execution engine. It just comes "for free."
I smell Josh Shinavier.
Marko.
http://markorodriguez.com
