Research On Semantic-based Query Expansion For Entity Search

In the era of big data,social networks,mobile internet,Internet of Things and other fields generate huge amounts of data every day.The explosive growth of data leads to information overload and changes the users' search needs from the document to entity,resulting in the booming of entity search.Unlike document search,entity search aims to find specific entity objects from heterogeneous data and is stricter in search quality.Traditional query expansion methods can effectively improve the quality of document search,but there are some problems such as limited source of extensions,low efficiency of query expansion and query drift problem,which reduces the performance of query expansion and are not fully appropriate for entity search.To solve the above problems,a semantic query expansion method for entity search is proposed.In the indexing stage,it stores heterogeneous information and supports fast acquisition of related items by constructing hierarchical semantic index.In the query expansion stage,it uses different generation methods and selection strategies to obtain structured and unstructured extensions based on established semantic index,and then utilizes the association between extensions for extension optimization.Finally,a composite probability model is used to merge and sort different types of extended query results to get the final results.Combined with the features of extension,this method takes full advantage of the efficiency of semantic index and the advantages of different query expansion methods,and improves the effectiveness of query expansion and effectively avoid query drift problem while guaranteeing the efficiency of query expansion.To verify the performance of semantic-based query expansion method,performance tests were carried out on the dataset sampled from ClueWeb09 Category B.The experimental results show that the proposed semantic-based query expansion method can effectively improve the performance of entity search and alleviate query drift problem while guaranteeing query efficiency.