Abstract: Many research centers and medical institutions have been accumulating a vast amount of various biological and chemical data over the past decade and this trend continues. Based on Linked Data vision, many semantic applications for distributed access to these heterogeneous RDF (Resource Description Framework) data sources have been developed. Their improvements have brought about a decrease of intermediate results and optimizing query execution plans. But still many requests are unsuccessful and they time out without producing any answer. Also, the applications which operate over repositories taking into consideration their specificities and inter-connections are not available. In this paper, the SpecINT is proposed as a comprehensive hybrid framework for data integration and federation in semantic data query processing over repositories. The SpecINT framework represents a trade-off solution between automatic and user-guided approaches, since it can create queries which return relevant results, while not being dependent on human work. The innovativeness of the approach lays in the fact that the coordinates of graph eigenvectors are used for the automatic sub-queries joining over the most relevant data sources within repositories. In this way searching can be effected without a common ontology between resources. In experiments, we demonstrate the potential of our framework on a set of heterogeneous and distributed cheminformatics and bioinformatics data sources.
Keywords: Federated SPARQL query, data integration, matrix eigenvectors
Journal: Semantic Web, vol. Pre-press, no. Pre-press, pp. 1-19, 2018