Query-Centric Semantic Partitioning (SPARTI)

April 23, 2018 ยท View on GitHub

  • Adaptively partition based on query-workload
  • Precompute Bloom join between the most frequent triples joins (MF-TJ) combinations
  • Partition related properties based on a greedy algorithm and a cost model
  • Current version is implemented to run over Apache Spark