Thesis and Dissertation Workshop (WTDBD)
Information Quality, Linked Datasets, Semantic Web
The increasing availability of datasets on the Web of Data faces some new challenges. One of them regards how to evaluate the quality of these datasets, whose solution may use Information Quality criteria. Most quality criteria are abstract concepts and need to be calibrated within a good situation before being used to evaluate any task. In this context, this work proposes an approach to evaluate the quality of linked datasets, i.e., datasets published according to the principles of Linked Data. Furthermore, we intend to evaluate the quality considering the application´s requirements. One distinguishing issue of our approach is the use of an extensible repository of quality criteria, which can be defined by a domain expert of the application. As a result, a quality score is generated and a ranking of the linked datasets is produced. This classification intends to serve as a comparative metric for the selection of the best datasets to be used in the process of rewriting queries and data integration.
Walter Travassos Sarinho, Damires Souza, Bernadette Farias Lóscio