Research On Negative Link Prediction Algorithm Based On Sentiment Analysis In Social Network

With the rapid development of the online social network in recent years,more and more analysis and research has been focused on the relationship between users.Link prediction is using the existing information in the network to predict future links.Since positive relationships are exposed on social networks,negative relationships are hidden.Much of the current research focuses on the positive link prediction,but the importance of negative links is underestimated.Negative link prediction is using positive link and the interaction between users to predict the potential negative relationship between users.Existing negative link prediction algorithm mainly consider the number of negative interaction between users,which doesn't make full use of users' interactions.So,we proposed a negative link prediction algorithm based on sentiment analysis.Firstly,combining the sentiment analysis method and social network,proposes a sentiment polarity strength calculation method of social network text.Then proposes an algorithm of building user interaction relationship matrix based on users' interactions.Secondly,building a reliability weight matrix based on user interaction relationship matrix,which is used for measuring the reliability of negative links.Then proposes an optimization negative link prediction algorithm based on structural balance theory.We carry out an experiment on the real data set,and evaluate the effect of various parameters through controlling variable method.Then,comparing our proposed algorithm with existing prediction benchmark algorithms,which demonstrates the proposed algorithm in this paper offer a good performance.