Using Stylometry and Machine Learning...
summary
The talk covers using stylometry (text analysis to extract characteristic writing features) together with machine learning methods—neural networks, decision trees, and Bayesian learning—to identify authors of anonymous online messages and detect behavioral patterns of users involved in crimes like hate speech, coordinated attacks, or false bomb threats. It presents this forensic approach as useful when offenders use multi-layer anonymization that defeats conventional cybercrime investigations, and includes a case study demonstrating practical machine-learning identification of an anonymous post's author.
AI-written summary · about
chapters (7)
- 0:00Introduction
- 3:38Bomb threat alarms in Poland
- 8:15Stylometry and applications
- 15:39Analyzing anonymous texts
- 20:36Document normalization and comparison
- 37:48Analysis results and summary
- 40:00Q&A
related talks
06mCreepy stuff with anonymous data
45mApplying Machine Learning Methods to Static Analysis of Executables
44mApplied Machine Learning for Data Exfil and Other Fun Topics
21mMachine Learning and Cyber Security - Detecting malicious URLs in the haystack
29mscikit-learn, machine learning and cybercrime attribution
30mData Driven Security
30mMachine Learning Techniques for Class Imbalances & Adversaries
32mAI/Machine Learning: What is Actually Working in Cybersecurity?