Topic-based Federated Query Engine

The increasing availability of open datasets makes it useful the adoption of novel query engines able to ease the retrieval of desired information from this huge amount of data. This paper proposes a smart discovery system that provides query engines with transparent execution of SPARQL queries over a collection of distributed datasets. The system adopts a topic-based matching algorithm based on different selection strategies to identify the target datasets of a query and their endpoints. These strategies do not require a priori knowledge about datasets and are used to rewrite the initial query by exploiting the Service keyword of SPARQL to split the query among different target datasets. Existing benchmarks are used to test the proposed middleware and to compare the adopted strategies through performance indicators.