The Design And Implementation Of SPARQL Based Semantic Web Data Retrieval System

Semantic web technology has attracted much attention recently among IT companies, institutes and researchers as well internationally, with the development of Web-of-Data, more and more semantic web data files have emerged and keep growing. The sheer size of such a data file has confronted the efficiency problem of information retrieval, while this is the key point along its way into practice. As for a standard solution, W3C has published SPARQL in order to unify the interface implementation of different graph databases and frameworks. However, this uniform solution is not quite sufficient for a non-professional user due to its grammar specification complexity as well as its requirement of ontology-awareness.In order for a user-friendly interface to query the semantic web data, we have designed a system which supplies multiple choices including key-words driven method, natural language based approach and finally a graphical user interface for the user to edit query with. Our system first requires the user to input strings to describe its query intention, and then retrieves URI to construct formal SPARQL query, which is finally delivered to the databases to get results. The URI retrieval process plays the most important role in our system design, during which we get the canonical URI string within the ontology from the user input label string. We have proposed a transformation assisted, string similarity based top-K similar string retrieval method, our method would utilize the ontology information to weight the label strings while calculating their similarities with input string, experiments show that it improves the probability of user-preferred label being included in the top-K retrieval results.Finally, we developed a demo system to demonstrate the availability of our approach, any non-professional user could make use of it to query the semantic data handily.