Research On The Semantic Retrieval Method Over The RDF Graph Of Hierarchical Clusting

Through the continued efforts of the W3C organization, industry and academic, the Semantic Web forms the multilayer architecture with RDF as the core of it. RDF defines the description of data semantics which computer could understand with a particular expression. Because the special data structure of RDF is different from a relational database and a web page, it raises many new issues in information retrieval applications and gets more and more attention.As to the retrieval of an RDF graph, taking into account the simplicity of user operation, the keyword query is still the mainstream way. There are Complex semantic relationships between RDF data nodes, and the text description of the data node is shorter, so the traditional keyword search techniques can’t be simply applied to the retrieval of the RDF graph. Otherwise, not only will a direct result of the search results lacking of semantics seriously, but also reduces the retrieval recall rate and precision precision rate.In this paper, focus on the RDF graph structure and semantic feature, a hierarchical concept index of RDF graph is built, and then a semantic retrieval method based on concept index is proposed. Firstly, preprocessing the RDF graph, extracting the entities from RDF graph, concept expansion of entities and hierarchical clustering, then a hierarchical concept index used to the semantic retrieval is bulit. Secondly, expanding the user’s retrieval request to make it with semantic, getting a combined request which contains concepts, entities and keywords; mapping concepts to the semantic space proposed in this paper, taking it in the hierarchical concept index retrieval and getting back results set. Finally, using the first results set in RDF graph to make the second query and returning the final results to the user.The overall framework of the RDF graph hierarchical clustering of semantic retrieval is designed and a prototype system is implemented. Taking the data set of DBLP in the field of computer literature data in the prototype system for experimental verification, the results confirm the availability of semantic retrieval method of hierarchical clustering. By using the hierarchical concept index, the search range is narrowed, and retrieval efficiency is raised. Concept expansion of retrieval keywords and semantic space make the search results with more semantic features, thus the precision rate is improved:through the second search of the RDF graph,the recall rate is improved.