Fusion User Influence And Similarity Collaborative Filtering Recommendation Algorithm

Nowdays,we are in an era of information overload.Users do not know how to do when they face a lot of choices,now the personalized recommendation system to build a bridge from the user to the article and filter out the void value and vapidity information,it always provide the favorite items to user and solve the problem of information overload.Collaborative filtering algorithm is the most popular algorithm in personalized recommendation.The traditional collaborative filtering recommendation algorithm refers to all the purchase history information of users when recommending for a target user.However,to recommend for a target user,the more similar users has more valueable information than other users.In this paper,aim at the problem of low accuracy of user similarity calculation in collaborative filtering.The main contents as follows:Firstly,aiming at the missing data problem of the user scoring matrix in the traditional cooperative filtering recommendation algorithm,taking into account the different of user consumption time information has implicit impact in user similarity calculate,This paper presents a method to explore the influence of users' by timing information,This method can solve the problem of missing data,and improve the accuracy of user similarity calculation by this method.Secondly,taking into account the impact of user common scoring and popular items on user similarity calculations,bring in the popular goods weight and user common score weight factor,present a method mix influence factor of calculating the users similarity,this method can reduces the error of user similarity calculation and improves the accuracy of user similarity calculation.Once more,the user influence matrix and the user similarity matrix obtained from above two calculation methods to generate the user nearest neighbor set by the Top-N recommendation method,to collection the user's historical and to generate the final project recommendation set by Top-N,and then take shape the recommendation algorithm combines the user's influence and similarity,to improve the accuracy of the user's similarity calculation degree.Finally,The algorithm is simulated on the Movielens dataset and compared with some classical algorithms.