Alignment-based approximate SPARQL querying on linked open data

Yu Liu^1,2, Lei Chen¹, Shihong Chen¹

¹School of Computer Science and Technology, Wuhan University, Wuhan 430072, Hubei, China
²School of Computer Science and Technology, Wuhan University of Science and Technology, Wuhan 430065, China

With the growth of Linked Open Data, more and more applications are developed to take full advantage of its massive data. However, all these applications face an inevitable problem - how to retrieve information from these datasets with different schemas, which results in that a query for a dataset may get none answer from other datasets. To solve this problem, ontology alignment has been adopted in some Linked Open Data querying systems. In this paper, we follow this idea and make further efforts to find more approximate answers by employing relations and probability values in the result of ontology alignment. The fundamental of our method is the similarity between entities, which is used to evaluate the similarity of rewritten query relative to original query. In order to facilitate user to query other dataset with original query, an algorithm for alignment-based approximate querying is proposed. In experiments, the SPARQL queries for DBpedia are rewritten on the basis of alignment result between DBpedia and YAGO. The results of experiments show that alignment-based approximate querying can not only retrieve approximate results, but also overcome the problem caused by imprecise result of ontology alignment, which is very common for most of alignment techniques.