Research And Application On Content Recommendation Method Based On User Characteristics Data

With the rapid development of e-commerce in recent years,the problem of information overload is becoming more and more serious.In order to solve this problem,the major e-commerce sites are more and more dependent on the content recommendation system.Firstly,this paper introduces the present situation of research and application of content recommendation methods,and makes a research on the user characteristic data and data mining technology.Through the analysis and comparison of existing content recommendation technology,proposing improvement plan based on content recommendation method for user characteristic data to solve the problem of "cold start",data sparsity and "new interest" discovery in traditional collaborative filtering recommendation method.The cold start problem in collaborative filtering recommendation method,using the attributes of user data,mining user attributes similar to bring similar interests.First,according to the characteristics of new users,finding similar properties for new users.Then,according to the preferences of similar users,recommending products based on user attributes.For the problem of data sparseness,this paper analyzes multiple feature data to get the user's attributes,preferences and scoring features.Improve the similarity calculation in collaborative filtering recommendation method by combining the similarity of user attributes and the scoring similarity which is optimized.The optimization of scoring similarity is to fill the user project scoring matrix with the user's preference expressed by the user's characteristic data.Finally,according to the comprehensive similarity between users,can find the similar users for the target user,through collaborative filtering,to achieve the recommendation.Aiming at the problem of new interest discovery,this paper introduces the association rule data mining technology.By mining the frequent itemsets to find the relevant rules between the commodities are obtained.Combine the online feature data of the users to filter the results of the association rules and get recommendations.Finally,an e-commerce recommendation system is designed and implemented according to the recommended method proposed in this paper.And apply it to the e-commerce website for actual testing.Through the experimental analysis of the accuracy of the recommended method,the recall rate and the average absolute error,to prove that the improvement of the recommended method is effective.