Research On Query Expansion Method Based On Word2Vec

Due to the fact that users often cannot accurately describe their query needs when searching,or the input term is too short for the query,the retrieval system cannot accurately obtain the query intent,and in natural language,problems such as synonyms,synonyms,and ambiguities result in problems.The search system often cannot accurately and comprehensively return documents containing user requirements.Query expansion technology can effectively solve the problem of such word mismatch and incomplete query description.It can reduce the user query burden while improving the search efficiency.However,the traditional query expansion technology has the disadvantages that the user needs to change the query word to search or need to search the system for the first time to retrieve the result file and then perform secondary search on the analysis result.Therefore,the research on query expansion technology based on Word2 Vec is performed in this paper.The main work of this paper is as follows:First,for the problem of analyzing and re-searching the results of the initial search,this paper proposes a query expansion method based on Word2 Vec.This method uses the Word2 Vec tool to train the word vector set,which is based on the word vector.The similarity calculation between the user selects the similar words as the extended word candidate set,and then filters the words in the candidate set according to the similarity degree to build an extended word list,thereby avoiding the initial search and document analysis of the user query and reducing the The burden of search.Second,the method of screening candidate sets by the size of similarity of query words is called Word2 Vec query expansion method based on query words.The similarity of words in query expansion method for query words is corresponding to its query.The word does not take into account the similarity with the entire query.This article also proposes a query-oriented query vector method based on Word2 Vec.This method treats the entire query as a whole to generate a query vector,and then calculates the candidate set of words.The similarity of the query vector is used to filter the candidate set to establish an extended vocabulary.This method takes into account the similarity between the expanded word and the entire query when building the extended vocabulary,and can effectively improve the retrieval effect.In this paper,a detailed description of query-oriented query expansion method and query-oriented query expansion method is presented,and several query expansion methods based on Word2 Vec are implemented on the computer platform.