Section: New Results

Semantic integration of heterogeneous data

A large amount of data sources are publicly available in heterogeneous formats such as relational, RDF and JSON. These data sources can share information about common entities, which the users may want to query as a single dataset, possibly exploiting also a set of semantic constraints which serve as a common integration perspective. We proposed a new approach to query such integration of datasources in a global RDF graph using an RDFS ontology and user-specified entailment rules. Previous approaches to query answering in the presence of knowledge involve either the materialization of inferred data, or reformulation of the query; both approches have well-known drawbacks. We introduce a new way of query answering as a reduction to view-based answering in [11]. This approach avoids both materialization in the data and query reformulation.

We have also developed an RDF Schema reformulation algorithm taking into account the reasoning on the ontology. This algorithm reduces query answering on data in the presence of an ontology, to query evaluation (solely on the data). In particular, this reformulation algorithm can be used to speed up query answering in the integration system mentioned above.