Research On Search Term Recommendation Method Based On Dependency Syntactic Network

The deepening of information overload and the rise of cross research are putting forward higher requirements for the filtering ability of information retrieval system.Search term recommendation is a basic information retrieval service to reduce the impact of irrelevant information and improve information quality.This paper introduces the dependency syntax theory and language network theory in computational linguistics to reveal the relevance of scientific concepts in scientific literature,takes these relevance as a standard of information screening and filtering,and proposes a search term recommendation method based on language network.Based on 124516 literature abstracts in the field of Information Science ?Library Science on the web of science platform,this method constructs the search term set and dependency syntactic network,and uses the Page Rank algorithm as the search term ranking basis to recommend the search term.The research based on user feedback shows that the accuracy of search term recommendation method based on language network is 80%;The average cosine similarity in the recommended list is 0.53,and the average Jaccard similarity in the table is 0.39.The diversity of search words in the table is better;There is no significant correlation between the importance of search terms and cutting-edge rating;The performance of surprise degree is significantly improved compared with the method based on word similarity and word clustering.The search term recommendation method based on language network can effectively assist the formation of research content in actual retrieval cases.In addition,in the process of method construction,this paper also combs the relevant theoretical and methodological framework.Firstly,this paper puts forward the basic research framework of keyword recommendation,and discusses the meaning,object,goal,method and evaluation of keyword recommendation.Secondly,from the framework of search term recommendation method,the framework of search term recommendation method based on WOS?LIS language network.This paper combs the construction framework of the method from three levels of the implementation of the search term recommendation method in LIS network.These research results can provide theoretical reference for the research of other search term recommendation methods based on language network.