| Question asked
|
Apart from introducing a novel indexing me … Apart from introducing a novel indexing mechanism for RDF, this paper also introduces an extension to SPARQL, in the form of P-path edges. However, it's not clear why P-paths are necessary, nor why the authors chose to model them as they did (as sets of predicates, as opposed to, say, sequences of predicates, regular expressions, or other formalisms used in proposed path extensions to SPARQL). They obviously play no part in the average query times which the authors use to compare GRIN with Jena, Sesame and RDFBroker, as these tools only deal with standard SPARQL queries. Why are the 20 non-standard TAP and ChefMoz queries mentioned at all? Can anything be said about GRIN's performance for these queries? The authors claim that "GRINAnswer is *often faster* than Jena, Sesame and RDFBroker for *certain types* of graph-based queries", which seems like a weak statement, especially since we're not told which queries these are. Do the advantages of GRIN have more to do with efficient evaluation of standard SPARQL queries, or with the ability to evaluate path-based queries of the variety supported by GRIN? queries of the variety supported by GRIN?
|
