gtfierro/brick_database_eval

Brick Database Eval

DBs to evaluate:

• Jena/Fuseki:
  • also need to experiment w/ different ordering of clauses in SPARQL query
• alegrograph
  • has free download!
• rdf3x
• python rdflib
• rdf4j
  • previously called Sesame
• top braid
  • expensive!
• Cayley/gremlin:
  • supposedly full Gremlin query lang can translate btwn SPARQL, but the subset currently implemented in Cayley cannot
• neo4j/sparql:
  • blog w/ basic directions: http://www.snee.com/bobdc.blog/2014/01/storing-and-querying-rdf-in-ne.html
  • seems to not work any more because the maven repo for openrdf is gone
• blazegraph: has free download:
  • supports SPARQL/RDF
  • Previously called BigData

TODO:

• dgraph:
  • supports GraphQL. Is probably 'compatible' with SPARQL, but involves rewriting queries
• stardog?
• virtuoso
• sesame

Process

• have a family of queries we want to run:

  • start with the 8 application queries for the Brick paper!

• need to "rewrite" the query for each database:

  • HodDB and rdf3x may need some tweaking

• 10 runs; take the average

• for each database, we have a directory containing:

  • Dockerfile for running the server
  • instructions for how to execute a query
  • code for running the queries

• which file do we run this on?

  • start with berkeley.ttl/soda_brick.ttl

Literature

• NOSQL database for RDF: an empirical evaluation
  • website: http://ribs.csres.utexas.edu/nosqlrdf/