Research On Method Of Fuzzy SPARQL Query On RDF

The Semantic Web has emerged as the expansion of the current Web and has been a hot research in the data and knowledge engineering field. It needs new data model to support unified access to Web information source, Web service and intelligent applications. RDF, as a data model representing network resources object and the relations between them, defines the underlying model that can be exchanged between applications without loss of information semantic.The main RDF query languages adopted in Semantic Web consist of RDQL, RQL, SeRQL, N3QL, Triple, and SPARQL. In real-world applications, there exists uncertainty and vagueness in user querying intention. However, most of the existing query systems on RDF can not process queries with fuzzy information directly and thus can not adequately meet users’ querying needs. Consequently, it becomes an important subject of Semantic Web to do research about fuzzy query on RDF. There are some drawbacks in current researches about fuzzy query on RDF. They cannot meet users’basic requirements of fuzzy query expressions and personal preference on returned results. Classical query system on RDF returns unordered result set and it does not satisfy user ranking demands.The core content of this paper is as following. On the basis of classical query language on RDF, SPARQL, the paper proposes a kind of fuzzy query method with flexible ability to represent fuzzy semantic information. Firstly, this paper divides fuzzy representation forms into two kinds, fuzzy terms and fuzzy relation. Furthermore, by combining the fuzzy set theory developed in recent years, an expansion part in FILTER constraints is got. Secondly, this paper comes up with a multi-dimensional preference structure, which provides user with flexible representation. In addition, using classical sort algorithm, this paper gives the ranking strategy combining preference and fuzzy membership degree. Besides, due to the application of crisp SPARQL implementations in fuzzy SPARQL syntax test, the work of this paper also covers defining the translation rules from fuzzy queries into crisp ones. Lastly, this paper gives the fuzzy query prototype system, shows the design of experiment and conducts data analysis of the experiments. The result shows that the proposed query method with orderly linguistic value and multi-dimensional preference structure is of high recall ratio; meanwhile the method gives a strong advantage in terms of precision, owing to the similarity of fuzzy SPARQL syntax with human natural language. Moreover, the implemented fuzzy query system has a good performance on system response times.