Research Of Ranking Based Recommender Algorithms

With the coming of big data era,the problem of "information overload" has been serious increasingly.Search engines can help people look for some information by retrieving keywords.It is hard to satisfy people's needs by this way.Each user has different purpose.However,recommendation systems can recommend something which people may be Interested in from massive information.Recommendation systems have been widely used in many industries,such as Amazon,Douban and NetEase Cloud Music.This paper proposed ranking based collaborative filtering algorithms.First we define users' preference of two films with product of degree and popularity.Secondly,we obtain the similarity between different audiences by cosine formula.Last we can get recommendation list by greedy algorithm.In order to compare with the commonly used collaborative filtering algorithm and the algorithm in this paper,we did multiple experiments.The results indicate that this algorithm have Superior performance on index of accuracy and NDCG.