Research On Personalized Recommendation System Based On Web Data Mining

With the continuous expansion of the scale of development of the internet and e-commerce, information overload is becoming increasingly serious. Personalized recommendation system is an important way to solve the problem of information overload, it can take the initiative to recommend to information which users are intrested in and dynamically make adjustments according to user needs. Now Personalized recommendation system has become an important part of e-commerce system.This paper has a research on the technology of data mining and web data mining and analyze the shortcomings of personalized recommendation system at first. Focus on the problems of cold-starting and sparsity, the system can rating the user’s favorite,browsing or associated goods in an implicit way by using web data mining and propose a hybrid collaborative filtering method including the method of K-means clustering and SVD. Through the experiment, the improved algorithm has better results in sparsity, accuracy and cold-starting. Focus on the problems of monotonous recommendation, this paper propose a function of the data weighting method based on time forgetting to reflect the change of user’s interest. The weighted data in line with the people’s forgotten laws and recommendation algorithm will make adjustments for the change of user’s interest in time. To a certain extent, it solve the problem of monotonous recommendation. Finally, personalized recommendation system based on web data mining have been designed by multiple perspectives such as the function, structure and so on, and make a recommendation engine improvements.