Analysis of fraud and money laundering in cryptocurrency

Abstract

The emergence of cryptocurrency and its blockchain technology has seen great progress and more widespread acknowledgment of its value as a digital currency. Cryptocurrency invites more criminals to carry out their illicit activities due to the pseudonymity it provides. Forensic analysis of such criminal activities like money-laundering generate new insights for a better safeguard of the financial system. Advances in machine learning algorithms motivates such opportunities. In this project, the author achieved the task of visualising the Bitcoin network on the Neo4j graph database for an accurate depiction and the explanation of each components that make up the cryptocurrency. Following that, the project explored on experimental results of a binary classification task in predicting illicit transactions using machine learning models like Logistic Regression, Random Forest and Multilayer Perceptron, with a discussion of using new methods such as Graph Convolutional Network to better utilise the relational information that Bitcoin has to offer.