Analysis and modelling of semantic attacks in online social networks

Abstract

The emergence of online social networks as an important media for communication and information dissemination during the last decade has also seen the increase in abuse of the media to spread misinformation, disinformation and propaganda. Detecting different types of semantic attacks possible in online social networks would require their accurate classification. Drawing similarities with other social computing systems like recommender systems, we propose a new taxonomy for semantic attacks in online social networks. We propose an algorithm which uses social network as a medium of social computing to analyse patterns of propagation of information and identify sources of disinformation in them. We construct a new information repropagation graph from social network data and carry out iterative core analysis of the graph to isolate possible contents of misinformation and user nodes which are involved in their propagation. We used seven different datasets obtained from Twitter to validate our results. We also propose an information propagation model for deliberate spread of false information in online social networks.