Challenges of using machine learning algorithms for cybersecurity: A study of threat-classification models applied to social media communication data

This chapter is aimed to introduce how researchers and security experts are using forums and social media posts as source for predicting security-related events against computational assets. First, we present an overview of the data-driven approaches and methods for processing the natural language communication extracted from the mentioned sources. Afterwards, we present an overview of common activities that take place on these channels; for instance, the trade of hacking tools and the disclosure of software flaws (vulnerabilities) on social media online communities. In the end, the chapter concludes with a discussion regarding the challenges of using learning-based techniques in cybersecurity. Moreover, we highlight some unsolved problems regarding the ethical considerations of using the content of hacker online communities for cybersecurity research.