Advancing Drug Dealing Detection Using Neural Embedding and Nearest Neighbour Searching Techniques

Abstract

The proliferation of the internet has also led to an increase in the actions of bad actors, such as those who sell drugs and pornography online. These illegal operations are made easier by the internet's relative anonymity, which makes it harder for platforms to properly enforce their policies. There is a crucial knowledge vacuum about how users create new accounts to get around bans, despite a lot of research being done to identify these bad actors. By focusing on the identification of drug dealers who avoid current regulations and detection systems, this paper aims to close this gap by developing a neural embedding and a nearest-neighbor search mechanism. The research attempts to identify the strategies these offenders use to carry out their illegal actions despite being prohibited by examining patterns of behavior linked to them. This project aims to develop robust detection systems that can identify networks of accounts that share similar attributes indicative of drug sales. The results may offer law enforcement and internet platforms useful information that will help them battle online drug trafficking and improve user safety through the implementation of more effective measures. In the end, this research advances our knowledge of the dynamics of online crime and helps to create proactive and preventative measures.