Enhancing Smart Contract Security: A Machine Learning Framework Using Natural Language Processing and Unsupervised Techniques

Dissertation presented as the partial requirement for obtaining a Master's degree in Data Science and Advanced Analytics, specialization in Business Analytics ; This thesis explored smart contract security using natural language processing and unsupervised machine learning techniques. By analyzing reports from Code4Arena and applying k-means and Latent Dirichlet Allocation, I aimed to identify trends in smart contract usage, vulnerabilities, and potential attack methods. My analysis yielded valuable insights for blockchain developers and cybersecurity professionals. The research identified trends in smart contract security threats and the potential of using machine learning for vulnerability detection. I propose a framework to enhance smart contract security based on real-world case studies and data from various blockchain platforms. This framework, informed by the identified trends, can contribute to building more secure blockchain ecosystems. The results indicate that the developed pipeline can efficiently evaluate various smart contracts, uncovering new vulnerabilities and attack types using a severity score.