A Novel Framework of Identifying Chinese Jargons for Telegram Underground Markets

As one of the most popular instant messaging (IM) software, Telegram has reached 500 million monthly active users (MAU) up to January 2021. Nevertheless, the characteristics of safety and openness have also made it a popular platform for transactions in underground markets. Moreover, cybercriminals usually use jargons instead of sensitive terms when they communicate in Telegram groups. Nevertheless, jargons identification relies on time-consuming and lagging manual work currently. To solve this problem, this paper proposes a Chinese Jargons Identification Framework (CJI-Framework) to identify jargons automatically. Firstly, we collect chat history from targeted Telegram groups to build a corpus called TUMCC, which is the first Chinese corpus in jargons identification field. Secondly, we extract seven brand-new features which can be classified into three categories: Vectors-based Features (VF), Lexical analysis-based Features (LF), and Dictionary analysis-based Features (DF), to distinguish between Chinese jargons and commonly-used words. Furthermore, we use a word vectors projection method and a transfer learning method to improve the quality of word vectors generated from the corpora. In our experiments, the CJI-Framework reaches a remarkable jargons identification performance with an F1-score of 89.66%. This work provides a method of identifying Chinese jargons for Telegram underground markets effectively and will be helpful for cybercrime investigation. It can also be helpful to jargons identification related to other similar communication platforms and languages.