Simulation of Personalized Recommendation Algorithm for User Feedback Data of Social Networks
Generally,social networks involve large numbers of users and content,so the computational complexity of personalized recommendations for all users and content is very high,and it is difficult for recommendation systems to capture users'preferences and interests.Therefore,this paper proposed a personalized recommendation algorithm for user feedback data in social network.Firstly,we calculated the similarity between time and space,and combined it with the trust relationship between users to accurately capture the personalized information of the target user.Second-ly,we combined the global similarity method with the similarity of user neighbors and item neighbors to build a matrix decomposition model for global similarity calculation.Then,we adopted the clustering algorithm to provide personalized recommendations based on user clustering characteristics.Finally,we constructed an interest graph and calculated the participation relationship between users and clustering topics.Meanwhile,we calculated the mutual dis-tance and difference index between personalized vectors.Thus,we achieved accurate personalized recommendations.Experimental results show that the confusion and average cosine similarity indicators are satisfactory,indicating that the method can provide precise and similar recommendation results that are similar to user interests under the same condition of personalized interest recommendation.
Social networksUser feedback dataPersonalized recommendationActivity areaSimilarity calcula-tion
万方数据
维普
