A Kind Of Collaborative Filtering Recommendation Algorithm Based On Social Networks

With the rapid development of large data era and Internet technology,it is difficult to obtain accurate information about the content they are interested in easy access to information at the same time.To solve the problem of information overload,many scientists and engineers proposed a variety of methods to collaborative filtering,represented personalized recommendation technology is an important means to solve the information overload,and has been widely used in many fields.However,the existence of collaborative filtering sparsity,scalability issues and cold start problems.On the basis of in-depth analysis of recommendation algorithm at home and abroad,focusing on collaborative filtering techniques were studied,we propose two algorithms to solve some problems exist in collaborative filtering algorithm.First,collaborative filtering recommendation algorithm based on user for analysis algorithm deficiency in similarity calculation,a new user is added in the similarity calculations and project co-Rating Regulators cold popular collaborative filtering algorithm AD-UBCF(Advance User-based Collaborative Filtering)algorithm.Regulators cold hot item is an adjustment attribute for the project itself,the popular project allowed to reduce the weight,the weight of popular items allowed to rise.User Rating is a common adjustment factor for a similarity calculation between users,user score more common,the greater the value adjustment factor,the less the value of the common score,the smaller the value adjustment factor.Finally,experimental verification of adding these two regulators can improve the recommendation accuracy.Secondly,due to the rise of social networking sites,traditional collaborative filtering does not consider social factors on the recommendation result,effective use of social networking can not only improve the quality of recommendation also improve the user experience,the user's social buddy relationship reflects to some extent they are interested in on similar interests,the more closely the relationship between people in a social network similar interests in their higher probability,therefore,on the traditional collaborative filtering based on the similarity calculation proposed adding a user importance of integration Social networking recommendation algorithm,namely SC-UBCF(Social User-based Collaborative Filtering)algorithm,importance of users is calculated based on the user's social relations,improve the accuracy of similarity calculated by the user's social network friends information,thereby improve the quality of recommendation.Finally,design three experiments by using BaiDu movie datasets validate the algorithm.The first set of experiments to determine the number of common user ratings regulators optimal value of the parameter.The second set of experiments,and a third set of experiments,the average absolute error MAE results as the evaluation index,AD-UBCF algorithms proposed and SC-UBCF algorithm with the traditional collaborative filtering algorithm based on the recommendation of the accuracy of users were compared,show AD-UBCF algorithm and SC-UBCF proposed algorithm especially SC-UBCF algorithm is more excellent than the traditional user-based collaborative filtering recommendation algorithm performance...