Research On Personalized Query Based On User Behavior

With the rapid development of the mobile Internet,more and more people use the Internet to obtain information.The information on the Internet also increases rapidly with the increase in the number of users,and the speed of information growth is also accelerated.From the massive amount of information,obtaining the information required by users has become a hotspot in the field of information retrieval,and due to the richness of mobile phone applications,many users obtain the required information directly in the APP instead of through a web browser.The content on the APP cannot be crawled by search engines.For most APPs,the search method of the site search is based on keywords to match the results.This method often finds a large number of results,which is difficult to satisfy users and be personalized.Query technology is a way to optimize queries.The current personalized query has developed to some extent.Based on the analysis of the shortcomings of the previous retrieval algorithms,this paper proposes a personalized query algorithm based on user behavior and a personalized pseudo-feedback related query expansion method.Examples were carried out related experiments.The experimental results show that the two algorithms proposed in this paper significantly improve the query results.The main work of the paper is as follows:First,this article crawls the information of Meituan takeaway stores and products in Nanchang through a web crawler,constructs a similarity feature matrix between stores in units of stores,uses the k-means algorithm to cluster stores,and verifies Based on the similarity between store similarity and TF-IDF feature clustering,the experimental results show that the clustering effect of special diagnosis matrix based on the similarity between stores is better.On this basis,this article uses the user's usage records to build the user's preference model.This step is a key step in personalized query.Secondly,based on the user preference model,a personalized query algorithm is built.By calculating the relationship between user preferences and the correlation between query terms and stores,the two are combined to reorder the query results,and the top N The store is returned to the user.And build a virtual platform for take-out orders,and carried out relevant experiments on this platform.Compared with the basic query method and the query method based on TF-IDF,the experiment shows that the personalized query algorithm proposed in this paper is better than the other two methods.Finally,based on the personalized query method,this paper proposes a personalized pseudo-relevant feedback interpolation expansion method.In this method,the personalized initial test result is used as the source of the expansion word,combined with the original query word to perform a secondary query,and the result of the secondary query is returned to the user as the final query result.And conduct relevant experiments on the virtual takeaway platform.Experiments show that the effect of personalized pseudo-relevance feedback query is better than that of non-personalized pseudo-relevance feedback query.