Design And Implementation Of Personalized Recommendation Algorithm Based On Social Network

With the rapid development of computer networks,our society is undergoing tremendous changes.With the vast amount of information we now have access to through things like online shopping and search engines,this information overload presents many challenges.To alleviate this situation,recommender systems serve as a bridge between data providers and users,which can help users better discover and use information resources.The main task of the recommendation system is to use the user’s past preference data to predict its possible future interests,and recommend corresponding information resources to the user on this basis.At the same time,with the rise of social media,the social relationship between users has become closer and closer,which provides new opportunities for recommender systems to introduce social network information.By considering users’ social attributes,recommender systems can better deal with data sparsity and improve the reliability of recommendation results.Therefore,using user social attributes to improve the performance of recommender systems has become an important research direction for personalized recommender systems.Based on this background,this topic researches and explores personalized recommendation algorithms based on social networks,mainly including the following three aspects:(1)A personalized recommendation algorithm based on social network subnet structure information is proposed.First,the algorithm divides the raw data into three views:"user-item" view,"social friends"view and "co-purchase" view,describing users’ social and purchasing information.Then,by using the generative confrontation network,the subnetwork structure information of users in different networks is modeled.Finally,the algorithm integrates structural and semantic information into a unified recommendation framework to describe how users’ hobbies are influenced by their social connections.Experiments have proved that this algorithm can be efficiently applied to a personalized recommendation model based on social media.(2)A personalized recommendation algorithm based on directional information in social networks is proposed.First,retain the directional information generated by user attention in the social network as the original data input;second,use the generative confrontation network to model the directional attention information in the user-user social graph,through two generators and a discriminator To generate the feature representation of the user in the social network;finally,the obtained user feature representation is integrated into the recommendation algorithm as the social influence received by the user.Experimental results verify the effectiveness and accuracy of the algorithm on existing data.(3)Design and implement a movie recommendation system.The system implements the recommendation of relevant modules and conducts functional and performance tests to verify the effectiveness of the system and related algorithms.