Research On Algorithms Of Personalized Recommendation System

The rapid development of Internet is leading us into the era of the network.We are enjoying various advantages from abundant resource of information in the era,whereas are always overwhelmed with huge amounts of data at the same time.Because we often cannot get what we need in this huge data.Under the circumstances,personalized recommendation system appeared in time,and has been applied in plenty of fields with the development of the network.According to the needs of individuals or their preference,personalized recommendation system can make a personalized recommendation.Quality of recommendations rely on a variety of personalized recommendation technology,among which collaborative filtering(CF)is one of the most representative technology.Collaborative filtering has achieved great success for perfectly meet the needs of the people,however,it also confronts many problems and challenges in the process of development.Data sparseness and scalability are two representative problems to be solved at present.Based on in-depth study of collaborative filtering,this thesis mainly aims to research and improve one representative algorithms,namely item-based collaborative filtering algorithm.For data sparseness problem,it introduces the degree of difference and time function to improve the calculation method of similarity;with regard to scalability problem,it introduces fuzzy clustering and user preferences cluster,making reduction of calculated amount by decreasing the number of nearest neighbors search.Both methods were tested by experiments.The former improved algorithm is confirmed to effectively alleviate the problem of sparsity,and the latter one played a good role in reducing the number of nearest neighbors search and it obtained the range of parameter in fuzzy clustering.Nowadays,it seems like development of the network would never stop advancing,thus we will be more and more dependent on the personalized recommendation system.Therefore,the research of collaborative filtering recommendation algorithm and other personalized recommendation algorithm is not only urgent and important,but also has the profound significance for the long-term development of the network.