Consistency Evaluation of RDF Data: How Data and Updates are Relevant

Trust and quality maintenance have always been problematic in the Semantic Web RDF bases. Numerous propositions to address these problems of data integration have been made, either based on ontologies or on additional metadata. However ontologies suffer from a adaptation speed slower than the data evolution speed and metadata requires ad-hoc manipulations of data by addition of extra-data. In this article we propose an original approach, based exclusively on data from the base, to evaluate the consistency of a candidate update to a RDF base, and finally to know if this update is relevant to the base. Our approach is inspired by case-based reasoning and uses similarity evaluation and query relaxation methods to compare a candidate update to the data from the base. If the modifications of a candidate update make the target part of the base more similar to other part (s) of the base, then this candidate update is considered consistent with the base and can be applied.