Design And Implementation Of Personalized Recommendation System Based On Social Network

Due to the constant update and iteration of network teahnology,people are entering the age of information overload.In order to solve the problem,many excellent solutions were raised by scientists and engineers through a lot of research and practice,recommendation system is the most representative solution.By exploring the relationship with the user and the product,the recommendation system provides users with personalized service.As a hot direction,recommendation system has been widely used in some fieldsHowever,due to the rapid development of the Internet,the volume of data increases dramatically,which makes the traditional recommendation system no longer meet the requirements.The traditional recommendation algorithm has some problems,such as cold start,score sparseness,etc.,which leads to the bad recommendation quality.At the same time,it also puts a strain on the system's computing power,the performance of the system is getting worse and worse,which can not provide users with real-time personalized product recommendation services In addition,the traditional recommendation algorithm only considers the direct social relationship between users,but ignores the impact of user scoring preference and community clustering characteristics on recommendation quality.In view of the above situation,the research content of this paper is as follows(1)By combining social network community partitioning technology with collaborative filtering recommendation algorithm,a recommendation algorithm-FBRA based on improved fuzzy clustering community partitioning is proposed.Firstly,according to the user similarity matrix and the shortest path between users,the user relationship matrix is constructed.Then,we use improved FCM to divide social networking communities and form the user's nearest neighbor set,which effectively alleviates the problem of data sparsity and cold start.Finally,the neighborhood users' scores are used to predict and score the items that are not scored by the target users and generate the Top-N recommendation list.Experiments verify the accuracy and efficiency of the proposed algorithm.(2)A collaborative filtering recommendation algorithm PKM-UserCF based on user rating preferences clustering is proposed.First of all,in order to select the initial clustering center of the K-Medoids efficiently,we using the parallel computings characteristic of the MapReduce.The use the Silhouette Coefficient to determine the optimal clustering number K.By doing that we can avoid the problem of low clustering quality caused by the uncertainty of K value and unreasonable initial clustering center.Then,aiming at the large-scale social network data,we use MapReduce computing framework to replace the cluster center in parallel and design the improved k-medoids algorithm in parallel.What's more,according to the resource diffusion on the user item scoring binary network,the user's scoring preference is obtained,combined with the user similarity,we can reduce the selection from whole social network to nearset neighbor set.Finally,the Top-N recommendation for the target user is realized based on the user collaborative filtering algorithm.Experimental results show that the proposed algorithm is accurate and efficient(3)In this paper,a personalized movie recommendation system is designed based on Hadoop distributed framework and the proposed recommendation algorithm.The system has the following functions:?.The system can provide users with personalized movie recommendation service according to their historical behavior,and users can search and score any movie they want online;?.The system will push users the list of recent popular movies and annual high-quality movies in various countries and regions;?.The system has the characteristics of social network,and users can click Interest and hobbies pay attention to each other and become friends.Friends can share,forward,like and comment on each other.?.The system has community function,users can join the community they are interested in,and users in the same community have post,comment,like and other functions.