Topic-based Indexing of Federated Datasets

The increasing availability of open datasets calls for the adoption of novel query engines able to ease the retrieval of desired information from this huge amount of data. This paper proposes an in depth analysis of a topic-based indexing solution for building transparent, federated query engines for SPARQL. For the index construction, we assume the retrieval of RDF fragments from the distributed datasets while user queries are re-written by the engine with the support of the SERVICE keyword, recently introduced in SPARQL for distributed queries. To perform rewriting, we propose different strategies for discovering and selecting the dataset endpoints. The strategies are analyzed and compared throughout the paper by exploiting some benchmarks. The tests regard both the effectiveness and performance of the proposed strategies. The results show significant improvements if compared with other techniques.