Research On Algorithm In Personalized Recommendation Which Based On Social Network

As it comes to the age of mobile Internet,more people are able to create enormous data,and surf the Internet is becoming more convenient.At the same time,the users are facing enormous data in daily life.With the increasing of Internet information,more users are becoming confused when facing this condition.Which is called"Information Overload".So the personalized recommendation system is born at the right moment.It recommends different information that the users may be interested in to different users,based on the consequence of data analyzing.On the other side,social networking is playing a more important role today.People's interests may be affected easily by others,which provide much enlightment to recommending system.Recommending algorithm is the core of the personalized recommending system.User-based collaborative filter is one of the commonly used algorithms.But it still has problems of data sparsity.Some researches began to use social network to deal with data sparsity.But few of them pay attention to multilayer social networking and users'social behavior.In this paper,the author proposed a new method to use in recommending system.Based on the traditional user-based collaborative filter,the new algorithm took advantage of multilayer social networking and social behavior with and use optimized K-means clustering(OKC)algorithm to filter relationship that not in same preference.At the end of this research,the author made experiment on Baidu movie dataset to recommend movies to users in this dataset.The result indicates that the new recommending algorithm can relieve the issue in data sparsity and the accuracy is better than user-based collaborative filter with social network.