In social networks, the most studies focus on the trust prediction, but distrust cannot get enough attention. The distinct characteristics of distrust relations present challenges to traditional relation prediction. Distrust relations are very sparse in social network, and negative interaction data is too little. We embark on the problem to investigate the distrust prediction with only network topology. After achieving seven social distrusting-inducing factors, we adopt machine learning and optimization methods to model the prediction process. The framework of Distrust prediction in Signed social network (DP-SSN) is proposed, which can predict distrust relations without any interaction data. Empirically, we perform extensive experiments on real-world data to corroborate the effectiveness of the proposed framework.
Signed networkMachine learningData miningSocial networks
SHEN Pengfei、LIU Shufen、HAN Lu
展开 >
College of Computer Science and Technology, Jilin University, Changchun 130012, China
This work was supported by the National Natural Science Foundation of ChinaNational Key Technology Research and Development Program of China
NETL
NSTL
万方数据
