Research On Method Of User Preference Analysis Based On Service Experience Perception

With the rapid development of Internet technology,it is increasingly common for users to use applications for service consumption on mobile terminals.In the process of service consumption,users' emotional experience is a direct reflection of whether they are satisfied or not.How to accurately identify and analyze users' emotional experience,so as to judge users' service experience when consuming services,and to explore the factors affecting users' service experience is an urgent problem to be solved.On the other hand,users often choose consumption objects according to their personal preferences in the process of service consumption.User preference is implicitly reflected in user service consumption experience and user consumption behavior data.Corresponding algorithms should be designed to analyze user preference and accurately recommend consumption objects,so as to improve user service experience when consuming.This article conducts research based on the relevant data collected by users when they consume service on ele APP.Firstly,a data collector application is developed to collect and store text data and user expression image data during user service evaluation,then the concept definition and attribute division of user service experience are carried out,we use the data of the research field in this article to verify the division of attributes.When analyzing the overall service experience of users,to solve the problem of low accuracy in analyzing users' emotional experience solely relying on text data,we adopt the method of fusing text data and user expression image data of two different modes to accurately identify users' emotional experience Then we explore the factors that affect users' emotional experience of service based on the LDA model.Based on the analysis of user service experience,this artical uses feature-opinion pair mining algorithm to mine the topics of preference and emotional experience of users in text comment data,then we construct subject portraits with different dimensions by combining expression image data.Through the definition of user preference,this artical introduces a method based on dynamic time series to analyze and predict user preference.To solve the problem of data sparsity in the mainstream recommendation algorithms,this artical proposes a hybrid algorithm integrating multi-dimensional subject portraits for personalized recommendation.Finally,based on the above research results,we designed and developed a user preference analysis and recommendation application tool.It provides users with portrait modeling,user preference analysis and user recommendation services to guide users' consumption decisions and improve their service consumption experience.It is complicated to identify the service experience and recommend products when users use the application on mobile terminal.Therefore,different approaches are needed to solve the problems that arise in the research.In the data acquisition stage,this article develops a data collector application program to collect user data when user consume services.In the analysis of user service experience,this artical integrates different modal data to improve the identification and analysis of service experience.In the stage of user preference analysis and recommendation,a method based on dynamic time series is presented to analyze user preference,then we integrate multi-dimensional theme portraits to make personalized recommendations for users.Finally,we proved the effectiveness of the methods we designed through comparing experiments,and proved the research value and significance of this topic through application examples.