Research And Implementation Of Personalized Recommender System Based On Book Social E-Commerce

With the rapid development of E-commerce,the environment of book E-commerce is changing continually.On the one hand,the competition of the industry becomes more and more aggravated;on the other hand,the diversification of choices leads to shopping behavior of consumers tending to rational.Consumers tend to carefully recognize the difference launched by different Web site.The emergence of E-commerce integration of social networks brought new changes to traditional book E-commerce.New book E-commerce model combined with the advantages of social network,can improve user activity and increase user viscosity.Because of the large number of the books,consumers need to spend much time and energy to choose the books they want.In order to solve the problem of information overload during looking for books and satisfy consumers' personalized demand for books,personalized recommendation system arises.Traditional personalized recommendation method uses user behavior data to recommend products for users.However,book social E-commerce not only has the user behavior data,but also has friend resources.Applying the users' friend resources to the recommendation method and using friends' preferences to recommend books for users,is supplement to traditional recommendation method.On the basis of analysis above,the thesis designed and achieved the book social E-commerce system and focused on personalized book recommendation method.Moreover,in order to extend the users' social scope,friend recommendation algorithm based on book social E-commerce was researched.The main work of this thesis is as follows.(1)According to the related theory of social E-commerce,the overall scheme of book social E-commerce is designed.(2)This thesis analyzed the characteristics of collaborative filtering recommendation algorithm and improved the similarity calculation between items and nearest neighbor selection of item-based collaborative filtering.Through the research and comparison on improved item-based collaborative filtering recommendation and social-based recommendation,this thesis put forward a hybrid book recommendation algorithm.The experiment showed that the hybrid recommendation algorithm can effectively improve the accuracy and recall rate of the recommendation.(3)This thesis made a research on friend recommendation algorithm based on book social E-commerce and put forward a friend prediction algorithm which combined social similarity and interest similarity.Improved friend prediction algorithm considered both users' already existing social contact and interest.The experiment showed that the accuracy of the friend prediction was improved.(4)On the basis of the analysis and research above,this thesis realized book social E-commerce system and personalized recommendation system.Finally,system performance was tested.