Betrayed by the Keyboard: How What You Type Can Give You Away

### Forensic Stylometry and Author Attribution At 00:00, Matt Wixey (PwC Cyber Security) investigates forensic linguistics and computational stylometry, exploring how subconscious writing habits can attribute anonymous cyber threat actors across digital communications.

### The Interdisciplinary Foundations of Stylometry At 08:00, Wixey explains the crossover between corpus linguistics and digital forensics. Stylometry operates on the premise that every individual develops an idiosyncratic "idiolect"—a unique linguistic fingerprint shaped by education, regional dialects, and cognitive habits.

### Identifying Syntactic and Structural Markers At 16:00, the speaker details feature extraction techniques. Investigators analyze function word distributions, punctuation habits, capitalization quirks, paragraph structuring, and recurring spelling mistakes in forum posts, extortion emails, and source code comments.

### Controlled Honeypot Experimentation At 24:01, Wixey presents an empirical research experiment. By deploying vulnerable honeypot servers and seeding simulated target data, researchers baited offensive operators into submitting command histories and explanatory text, subsequently testing stylometric classifiers against the captured samples.

### Machine Learning Attribution and Evasion Countermeasures Concluding between 32:00 and 40:00, Wixey evaluates Support Vector Machines (SVM) and natural language processing classifiers for threat attribution. He also discusses counter-attribution techniques, showing how whistleblowers and defenders can obfuscate linguistic signatures using iterative machine translation and style-cloaking tools.

(AI summary — Gemini Flash)