Style and Identity Recognition

Knowledge of the author’s identity and style can by used in the fight against forged and and anonymous documents and illegal actions in the Internet. Nowadays, there are many systems dedicated to solving stylometric tasks, but they are predominantly designed only for a specific task; they are used exclusively by their owners; or they do not natively support any Slavic languages. Therefore, we present new open-source modular system Style & Identity Recognition (SIR). The system is designed to support any stylometric tasks with minimal efforts (or event by default) by combining dynamic stylometry features selection and prediction driven by input data labels. The system is free for non-commercial applications and easy to use, therefore it can be helpful for people dealing with threatening e-mails or sms, children forum protection against pedophiles and other tasks. Being customizable and freely accessible, it can be also used as a baseline for other systems solving stylometry tasks. System combines machine learning techniques and nature language processing tools. It is written in Python and it is dependent on other opensource Python libraries.