In recent years,knowledge base has attracted a lot of attention from academia and industry.As we all know that large knowledge bases like Freebase and DBpedia have become some important semantic databases which support an infinite domain Question answering system.However,manipulating the knowledge requires the abilities of writing and proficiency in semantic query language,such as in the Question answering system users should query the related data through the construction of SPARQL.However,most users are ordinary users who have no these abilities.So in order to solve this dilemma,we need to transform these natural languages input by the ordinary users into the SPARQL in the Question answering system to help them to search,so that the accurate answer can be insured.In addition,voice recognition service which can identify natural voice has become more and more popular,which is especially widely used in smart phones,this means that it is the time for us to transform the natural language question into structure query statements.In this paper,we mainly study the methods of transforming the natural language question into SPARQL structure queries through semantic query graph,and proposing a framework to help users to construct good structure query statements in knowledge system.Through extending and generating query semantics,first,this framework transforms the natural language question into the corresponding SPARQL structure queries,then query the RDF data set,finally the desired query results of users can be returned.The aspects of specific research results are as follows :(1)This paper transforms the natural language question into a SPARQL query by the middle form of the semantic query graph.In the paper the formal definition of semantic query graph and its construction method are given.The semantic query graph can not only express a simple question,but also a complex question and a compound question.So we can transform any types of questions into the corresponding SPARQL query by semantic query graph.(2)This paper proposes the overall framework of SPARQL structure query,which includes four modules,namely,the module of problem analysis,the module of semantic graph construction,the module of resource mapping and the module ofSPARQL generation.Besides the paper introduces the details of the specific functions and the operation process of each module.(3)According to the proposed SPARQL framework,through testing and comparing with other existing methods,the experimental results show that the proposed method can realize the transformation of the natural query to the structured query statement,which can reach an average of 0.825 by tested on the several data sets. |