关键词: RDF数据图,关键词查询,子图,相似度矩阵,统计语言模型

Abstract: As huge amounts of the semantic Web data have sprung up,people are more concerned about query efficiency over RDF data graph. Retrieving RDF data graph directly by keyword matching is an area of research focus.In this paper,a retrieval model was proposed,which enables keyword search for RDF graph.First,for the improvement of query efficiency,an algorithm named ISGR (an Iterative way to SubGraph Retrieval) was proposed,in which query keywords can be matched with subgraphs from RDF data graph,and a collection of subgraphs which should be unique and maximal is got.Next,in order to solve the problems of redundant results and deviation that frequently emerge in keyword search,a mixture ranking model(SimLM) was proposed,which considers the structural information between keyword graph and result graph,and mixs statistical language model.A numbers of contrast experiments over two kinds of open source real datasets prove that the retrieval and ranking model proposed in this paper outperforms well-known techniques in the field of consistency and relevance.

Key words: RDF data graph,Keyword search,Subgraph,Similarity matrix,Statistical language model