An Experimental Evaluation of Relational RDF Storage and Querying Techniques

MahmoudiNasab and Sakr

rdf triple store evaluation database performance

@inproceedings{mahmoudinasab:dsaa-2010,
  title={An Experimental Evaluation of Relational
         RDF Storage and Querying Techniques},
  author={MahmoudiNasab, H. and Sakr, S.},
  booktitle={Database Systems for Advanced Applications},
  pages={215--226},
  year={2010},
  organization={Springer}
}

Download PDF

SP^2Bench RDF querying performance benchmark

BenchmarX'10

Three primary classifications of relational RDF:

• Vertical (triple) table stores: Every triple is in the same 3-column table
• Property (n-ary) table stores: Properties are modeled as table columns
• Horizontal (binary) table stores: One table for each property

Metrics

• Loading time
• Storage cost
• Query performance

Yet another paper confirming that it's a toss up between triple store and vertical partitioning, with the latter suffering as the number of predicates goes up