Abstract : In this work, we present DRSS, a distributed and scalable engine for RDF streams processing. DRSS proposes a new query syntax for continuous querying of RDF data streams. The system includes among others three efficient algorithms for (1) rewriting continuous queries sharing common sub-structures (2), SPARQL query partitioning across multiple computer nodes according to an efficient distribution strategy and (3) query-based data distribution for local processing of sub-queries minimizing data exchanged across nodes. Our system combines both real-time data from multiple sources and stored RDF processing. DRSS and its all algorithms are implemented using the real-time data processing platform Storm Framework, which provides parallelization mechanisms of query operators. The DRSS evaluation is conducted on a real dataset containing up to 1 million RDF graphs. Experiments and obtained results confirm the scalability and the effectiveness of our system.
