Design And Implementation Of Movie Recommendation System Based On Social Network Analysis

In the era of rapid Internet development,movie data resources are growing exponentially.When faced with a huge amount of movie information,it is often difficult for users to find the information they need quickly and accurately,which leads to unsatisfied user needs and low usage of movie resources.The personalized recommendation algorithm helps users find valuable information by using the user's history and other supporting information to make recommendations.Current research on recommendation algorithms has made some progress,however,there is still room to improve the recommendation performance of the algorithms.With the development of social media,interactive relationships such as trust and attention are generated among users.Due to the shortage of traditional collaborative filtering recommendation algorithms,this paper demonstrates the effectiveness of the algorithm based on the community division of social networks and the role of social influence in recommendation,and using experiments on a classic movie dataset.The main research of this paper is as follows:(1)A recommendation algorithm based on community segmentation is proposed.First,the number of items jointly evaluated by users is added to the traditional similarity calculation method to calculate the similarity between users;Second,direct trust between users is calculated using user rating data,and indirect trust is obtained through the propagation of direct trust;Then,the user relationship matrix is constructed by combining the similarity and trust degree,and subsequently the user network is constructed based on the user relationship matrix,and the user network is divided into communities using the improved Louvain algorithm,and then the set of nearest neighbors is generated from the results of the community division;Finally,rating predictions are made for the items that have not yet been rated,and a recommendation list is generated.The experimental results demonstrate the effectiveness and accuracy of the algorithm.(2)A social recommendation algorithm based on attention model is proposed.First,modeling user,item attributes and social information using noise-reducing selfencoders and obtaining a low-dimensional vectorized representation of the three;Second,calculate the social influence of the user with all his social neighbors;Then,the attention model is used to generate information about the user's social context and to select influential social neighbors for the user;Finally,the user's rating of the items is obtained and the items that may be of interest to the target user are recommended from the rating list.The validity of the model is verified by comparing it with the benchmark model and analyzing the parameters.(3)Completed a movie recommendation system based on social network analysis.The system is based on the B/S design model and implements the functional modules related to movie recommendation from the user requirements.The implementation of the system solves the problem of movie resource overload and provides convenience for users to select movies.In summary,the algorithm proposed in this paper can effectively alleviate the data sparsity problem and improve the recommendation effect on movie datasets,while the movie recommendation system can find movies that may be interesting for users and save their search time.