Research On Keyword Query Approach Over RDF Data Based On Tree Template

With the development of semantic data and advancement of Linking Open Data, the opening RDF data continues to emerge and the demand for access to semantic data is also increasing. The problem that how to help users query data conveniently is one of the most concerned problems in current semantic web search. The fact that some search techniques which are based on keywords have been widely used in World Wide Web search engines suggests that keyword query has become a popular information search method. In this paper, the keyword query problem about large-scale RDF data has been conducted research. The work is as follows.(1) Propose an indexing method over RDF data based on tree template. Firstly, a large RDF data graph is divided into some D-step trees; Secondly, the content nodes as well as edge labels of D-step trees which are connected in breadth-first traversal order as index. As many D-step trees have the same structure, the edge labels are extracted from these D-step trees as tree template in this paper.(2) Design and implement a keyword query algorithm based on tree template. The algorithm makes use of the relationship between edge labels and content nodes of D-step trees, which can construct query results from D-step trees quickly. The experimental results show that the algorithm can provide rapid response to keyword query on large-scale RDF data.(3) Propose a new ranking method for query results. The method combines TF/IDF and PageRank to evaluate the query relevance between query results and keywords as well as the importance of the entities of query results. Finally, top-k most relevant results are returned to users.Besides, the above-described method in this paper has been experimented on DBpedia, and the experimental results show that the approach is effective.