Application And Research Of Behavioral Analysis Technology In User Management System

In recent years,users' demand for services has continuously shown a trend of diversification and individualization,thus providing users with a unified information service has been unable to meet users' individual needs.In the user management system,a large amount of user-related data are generated every day,which contain users' behavior information.Collecting and analyzing these data can help to discover users,behavior pattern,understand their real needs and provide personalized information services for users.The user management system contains a large amount of user data,so it is an important way to meet users' personalized demands and improve the system's service quality.by using related technologies for user behavior analysis and apply them in user segmentation and personalized recommendation.In this paper,related theories and techniques of user segmentation and personalized recommendation are studied,and contrast correlation algorithm,the K-means algorithm and collaborative filtering algorithm are selected to achieve user segmentation and personalized recommendation.Besides,for the deficiencies of the above algorithm Inadequacies,the following measures are proposed for improvement:(1)Aiming at the problem that in the K-means algorithm,the varying degrees of influence of clustering object's dimension attributes on the clustering result are not taken into account,so a K-means improved algorithm based on self-adaptive weight of discrete factor is designed.The improved algorithm can reduce the influence of unrelated dimensions on the clustering results and improve the clustering accuracy.(2)Aiming at the problem of user similarity calculation in user-based collaborative filtering recommendation algorithm,an improved Pearson similarity formula is designed.In the similarity calculation process considers two influencing factors are considered:users' common evaluation of the number of commodities and popularity of the product,making the calculation of the similarity between users more accurate,so as to obtain a better recommendation effect.(3)In practical applications,in order to improve the problem of user cold start and scalability in collaborative filtering recommendation algorithms,a hybrid recommendation technology based on clustering technology is proposed to personalize users in the system.Finally,this paper applies the above methods to the online medical system,designing and implementing the functions of user segmentation and personalized recommendation.The above methods are tested in the actual data,whose experimental results show that the improved k-means algorithm designed in this paper is applied to user segmentation,which can reduce the effect of irrelevant segmentation variables on segmentation results,and improve the accuracy of segmentation results.The improved collaborative filtering hybrid recommendation technology designed in this paper can improve the recommendation accuracy and the problems of cold starting and scalability for users in the recommended system.The validity of the method described in this paper is proved.