Caines, Andrew, et al. "Automatically identifying the function and intent of posts in underground forums." Crime Science 7.1 (2018): 19.

This study focuses on the automatic classification of posts from hacking-related online forums to better understand user behavior in cybercrime-related social networks. An annotation schema was developed to label posts by type (e.g., question, comment), author intent (e.g., positive, negative), and addressee (e.g., general audience or specific users). Manual annotation showed substantial agreement for post type and addressee, and fair agreement for author intent. Various classification models were trained, revealing that a hybrid logical-statistical model is most effective for post type and author intent, while a purely statistical model excels in identifying addressees. The findings have implications for analyzing forum conversations and identifying key actors in these networks.