Untrusted User Detection in Microblogs

As Micro blogs such as Sina Weibo and Twitter are becoming increasingly popular, a lot of users are benefiting from the convenience of social communication and information sharing. However, untrusted users can take advantage of these sites as gateways to make money, spread rumors and so on. This paper proposes a label propagation algorithm based on the trust model to detect the untrusted users on social networking sites. The algorithm consists of two main steps. Firstly, we conduct a hybrid graph with three types of nodes (users, tweets, topics) to model the trust relationship between each user pair. Secondly, we utilize a new method inspired by the Label Propagation Algorithm to detect more untrusted users based on a small known seed set. We conduct experiments with a real data set, and the evaluation demonstrates the effectiveness, efficiency, and scalability of the proposed solution.