Semantic Partitioning

Master Thesis from Bonn University

SANSA Semantic Partitioning is a scalable and highly efficient application that first perform in-memory RDF Data (N-Triples) Partition and then pass the partitioning data to the SPARQL Query Engine layer to get efficient results. It is built on top of SANSA-Stack using the Scala and Spark technologies.

Read here: Scala & Spark

Benchmarks

The datasets should be in N-Triples format.

LUBM

./generate.sh --quiet --timing -u 1 --format NTRIPLES --consolidate Maximal --threads 8

BSBM

./generate -fc -s nt -fn dataset_10MB -pc 100

DBpedia

direct download

Application Settings

VM Options

-DLogFilePath=/SANSA-Semantic-Partitioning/src/main/resources/log/console.log

Program Arguments

--input /SANSA-Semantic-Partitioning/src/main/resources/input/lubm/sample.nt
--queries /SANSA-Semantic-Partitioning/src/main/resources/queries/lubm/query-01.txt
--partitions /SANSA-Semantic-Partitioning/src/main/resources/output/partitioned-data/
--output /SANSA-Semantic-Partitioning/src/main/resources/output/query-result/

SPARQL Operators

Read here: SPARQL Operators

SPARQL Queries

Read here: SPARQL Queries

Deploy App on Cluster

Read here: App Deploy

Future Work

• Implement Prefix for SPARQL queries
• Add more operators
• Add support in FILTER:
  • Math: +, -, *, /
  • SPARQL Tests: bound
  • SPARQL Accessors: str
  • Other: sameTerm, langMatches, regex
• Show predicate in the final result (for flexibility)