Research On Personalized Search Method Based On Language Model

In recent years,under the wave of the Internet,the scale of information has reached an unprecedented level,presenting the trend of "information explosion".How to obtain information that meets the needs of different users from the massive data becomes a top priority,and it is also a problem that needs to be solved urgently in the field of information retrieval.In the field of information retrieval,the integration of user information and the construction of personalized search models that meet the needs of different users have become mainstream.Personalized search models often include methods based on user search history,search results,and search engine history records.Most of these methods are based on non-public information of users to build search models.However,the information is characterized by greater difficulty and noisy.Therefore,in the face of increasingly diverse user demands,a user's portrait can be constructed by using a popular social platform to describe the user's interest and preference.This paper uses the microblogging social platform to build a user interest model to meet the search needs of different users,and to explore the feasibility and effectiveness of personalized search methods.The main work is divided into the following two aspects:Firstly,this thesis uses user microblog information to build a user interest model and achieved personalized search method.The specific steps are as follows:To begin with,this thesis uses Scrapy framework to crawl microblog corpora,preprocesses corpora,builds the data set needed for the experiment,and builds a standard data set in the form of manual annotation.Then,under the framework of the language model,this thesis uses the user's microblogging data to build a user-based microblogging-based personalized search method(UPM).Subsequently,using the user's social relationships,a personalized retrieval method based on collaborative computing(CCUPM)was constructed.Finally,the UPM and CCUPM were compared with the baseline model.The experimental results verify the feasibility and effectiveness of the personalized retrieval method.On the basis of user interest modeling,this thesis integrates pseudo-related feedback to realize personalized retrieval.First of all,in the personalized search method,the initial search results are obtained and the resulting documents are regarded as feedback documents.Then,this thesis uses the feedback documents to achieve the expansion of the query,build a personalized retrieval method based on pseudo-related feedback(PRF-CCUPM).Finally,the PRF-CCUPM was compared with the baseline model.The experimental results verify the feasibility and effectiveness of PRF-CCUPM.Through the above two aspects of research and verification analysis,this paper successfully applies the user model to the field of information retrieval,and provides a feasible solution for the research in the field of information retrieval.