Research On RDF Fuzzy SPARQL Query Results Ranking Based On Contextual Preference

The Semantic Web has emerged as the expansion of the current Web, has been a hot research in the data and knowledge engineering field. The core idea of the semantic web is by adding some semantic information on the Web to achieve the improvement of the representation and access to information, to make the information on the Web can be understood by machines that makes retrieve information more efficiently and accurately. In Semantic Web, RDF and RDF Schema are commonly used to describe resource, so as to provide the semantic information to the resources, and SPARQL is a kind of RDF query language. As more and more information on the Web are described using the RDF language, research on the RDF query has become an important subject.In the RDF query, because of the fuzziness of the user query intention, they hope to be able to express the fuzzy query request, for this, the researchers have launched some studies, and have appeared some research rusults of the fuzzy expand to the SPARQL. As these reaearch results of the fuzzy query appear, it also brings some problems, because the fuzzy query request is loose, there will be a lot of results returned, and all of these results are in disorder, which make the user difficult to find the information they need. So users may wish to see the results which are the most meet their preferences first, so the rank of the fuzzy query results is particularly important. But the existing fuzzy query results ranking methods does not fully consider the preferences of the users.To resolve the problem of users hope the system can rank the fuzzy query results according their preferences, this paper studies the RDF fuzzy SPARQL query and ranking method based on contextual preference. First, a contextual preference model has been proposed, we use a simple quantitative preference model, by adding context in preference to strengthen the expression ability of the preference model. Then, we cluster preferences based on the sementic similarity of the context states, for each cluster generate a scoring table for all the triples in the RDF ontology to save storage space. At last, when processing users’ query, the revelent context states and triple scoring table are used to calculating the overall score of all search results, to quiekly provide ordered results for users. Expriment and analyses indicates that the method we propose is feasible. We also make some comparisons with the existing RDF fuzzy query methods and prove our method has stronger preference expressing power and higher precision.