Abstract—Influence maximization is the problem of finding a small set of seed nodes in the social community network and then maximizes the spread of influence under certain influence models. Under the current social network, influence has become an indispensable idea, and it has become the mainstream idea of people. How to expand their influence can be a problem worthy of discussion. Of course, the current influence not only includes positive influences, but also negative influences. There is an ancient Chinese saying that good things do not go out and the things in response to this statement go out, we use the negative effects that may occur in the network and minimize the negative effects of negative influences so that the positive impact can be maximized. Therefore, we introduce a quality factor q to change the negative influence and positively influence the influence of propagation in the network. We use our algorithm in the IC-N model and indicates that seed selection is related to the quality factor q. As the presenter of IC-N model shows that the model maintains a number of nice properties such as submodularity, which allows a greedy approximation algorithm for maximizing positive influence within a ratio of 1-1e. We also improve the way based on spreading paths. Our experiment results demonstrate the efficiency of our algorithm and show that our algorithm can spread more positive influence.
Index Terms—Social influence, influence maximization, negative and positive influence, independent cascade model.
Weijia Ju is with the Department of Computer Science, Yangzhou University, Yangzhou, 225127, China (e-mail: 1069025168@qq.com).
Ling Chen is with the State Key Lab of Novel Software Tech, Nanjing University, Nanjing, 210093, China (e-mail: yzulchen@163.com).
Cite: Weijia Ju and Ling Chen, "Influence Maximization in Social Network with Negative Influence," International Journal of Machine Learning and Computing vol. 9, no. 2, pp. 230-235, 2019.