Predicting User Influence in the Propagation of Toxic Information.

With the advances of information technology, the Internet has become an indispensable part of life. At the same time, toxic Information has become virulent and common on the Internet. Such information propagation can have a negative impact on individuals, organisations and the society. Traditional approaches, such as detecting texts and posts with toxic Information will eventually generate ‘dark pools in which the online propagation of toxic information will flourish. In this study, we pay attention to influential users who evidently affect others in the activities related to toxic information. A method of predicting user influence was proposed. Compared to the existing literature, user influence is assessed on the basis of users’ text-based and behaviors-based characteristics rather than the network structures only. Moreover, whether the influential users have always been those with strong connections on the social networking site is also discussed. The effectiveness of the proposed method is demonstrated in two real-world datasets.