Authorship Attribution on Crime Related Social Media
Research on the Darknet in Forensic Linguistics

Area 10 – Scienze dell'antichità, filologico-letterarie e storico-artistiche

Selecting prolific users of an organized crime–related online forum, this volume presents experimental research that attempts to find the author of anonymous texts. Firstly, the reader is introduced to two increasingly popular tasks in forensic linguistics: authorship attribution and author profiling. Secondly, the most common automated text classification techniques are presented in order to test their performance in authorship attribution. At the end, results show that the most successful experiments seem to capture unknown authors’ stylistic decisions or habits. This is accomplished using features that represent grammatically shallow and paralinguistic (not language, but rather language–related) information.

SINTESI