Using Text Mining Algorithm to Detect Gender Deception based on Malaysian Chat Room Lingo

Abstract

E-mail can be a fantasy playground for identity experimentations where players take on an imaginary persona and interact with each other in the virtual world. Therefore, gender deception is difficult, risky and it can be abandoned at will. Inference can be made both from writing style and from clues hidden in the posting data. A text-mining algorithm was designed to detect gender deception based on gender-preferential features at the word or clause level of Malaysian e-mail users. Based on this algorithm, a prototype in Visual Basic is developed. It was tested with 16 documents; each consists of five e-mails exchanges of respective individuals. The tests shown the prototype have 81.3% of accuracy level. This prototype can be a tool to assist interested parties such as the Criminology and Forensic Department, e-mail users interested parties such as the Criminology and Forensic Department, e-mail users and virtual communities to successfully identify gender deception.