Design And Implementation Of Personalized Shopping Recommendation System

Along with the rapid development of the Internet,information becomes more and more diverse,so it's harder for users to find the information they need.In order to solve the problem of “information overload”,the recommendation system is born.By analyzing the user's historical behavior,the recommendation system finds the user's interests,and then makes personalized information for this user.At present,the recommendation algorithms have been widely used in fields of e-commerce,music and social network,but there are still problems such as sparse data,cold start and so on.It has become a hot topic in the research of recommendation algorithm that how to solve these problems effectively.Based on analysis and comparison the existing recommendation algorithms,the thesis develops an improved multi-information incorporated recommendation algorithm,then designs and implements its application in practical recommendation system,aiming at solving the problems like data sparsity and cold start in the recommendation algorithm.The main contents are summarized as follows:1.Comparative study on the traditional recommendation algorithms shows that matrix factorization can provide more accurate recommendation results in sparse data set.Currently many recommendation algorithms integrate with social network information,but relative data is not easy to collect.So,this thesis brings in user attributes,the relationship of items and the time sequence behavior information into probabilistic matrix factorization,to develop an improved multi-information incorporated recommendation algorithm.2.Both theoretical analysis and experimental test on Movielens dataset,it is shown that multi-information incorporated recommendation algorithm is more accurate than other comparison algorithms,and it can effectively alleviate date sparsity and cold start problems.3.A comprehensive analysis of major e-commerce websites was conducted to determine the target of a personalized shopping recommendation system in this thesis.Design and implement a personalized shopping recommendation system,and illustrate the multi-information incorporated recommendation algorithm's realization in this system.The intelligent recommendation module of the personalized shopping recommendation system mainly includes: popular recommendation and personalized recommendation.The purpose of introducing hot recommendations is to mitigate cold start problem and provide users with more choices4.For the system requirements,a test program is formulated.A test environment is simulated in this thesis to test the system's function,recommendation algorithm and its performance.The results reveal that the application of the multi-information recommendation algorithm to a personalized shopping recommendation system is workable,and it can meet the expected functional and performance requirements.