Conclusion

In this book, we presented results of the efforts to detect “Pathogenic Social Media (PSM)” accounts who are responsible for manipulating public opinion and political events. There are many challenges in the area of PSM accounts detection. In Chaps. 3 and 4 , standard and time-decay probabilistic causal metrics were proposed to distinguish PSM from normal users within a short time around their activity. In Chap. 4 , we investigated whether or not causality scores of PSM users within same communities are higher than those across different communities. Furthermore, as available data for training automatic approaches for detecting PSM users are usually either highly imbalanced or comprise insufficient labeled data, in Chaps. 5 and 6 , we proposed semi-supervised approaches for detecting PSMs that utilize unlabeled data to compensate for the lack of sufficient labeled data. In Chap. 7 , we observed that PSMs would deploy techniques to generate diverse information to make their posts look more natural. We utilize several metrics to approximate the complexity and readability of content shared online by PSMs and normal users. Finally, in Chap. 7 , we took a closer look at the differences between malicious and normal behavior in terms of the posted URLs by different types of users. We leveraged several characteristics of URLs as source-level information along with other attributes in a supervised setting for detecting PSMs.