User Profile And Behavior Profile Based On Simulated Data

User portrait is a tool to describe user characteristics and needs,and the personalized recommendation system can provide users with recommendation services based on user portraits.Among them,the collaborative filtering algorithm is used most in the personalized recommendation system.However,the problem of sparse data and the problem of cold start of the scoring data has greatly affected the quality of its recommendation.Therefore,this article proposes a hybrid recommendation algorithm(UPICF)based on user portraits and improvement collaborative filtration.Among them,the user portrait consists of three aspects:(1)the natural portrait of the user;(2)the user’s interest portrait;(3)the user’s ability portrait.The user portrait combines the user’s basic information,the basic information of the product,and the scoring information,and then mixes with the collaborative filtering algorithm,which can make the algorithm more accurate recommendation results.In addition,the algorithm also improves the traditional collaborative filtering recommendation algorithm:(1)unifies the user scoring standards;(2)improve structure similarity;(3)improve value similarity;Degree;(5)Improve scoring prediction formula,so that not only can user portraits be used for personalized recommendations,but also can alleviate the sparse and cold start of user score data.Finally,weighted combination of recommendation algorithms based on user portraits and improvement collaborative filtering algorithms to obtain a hybrid recommendation algorithm,which has actual application value in the recommendation system.Because user portrait analysis will involve a large amount of user data,which also involves data security and privacy protection.Therefore,this article selects analog dataset for experiments and compares with other models.The experimental results show that compared with the reference model UPCF,the model proposed in this article increases the MAE index by 12.92%,and the F1 indicator has increased by 5.78%.The experimental results also show that the algorithm proposed in this article can effectively alleviate the scarcity of data and cold startup,and has a certain optimization effect.