Exploring vulnerabilities in verified ethereum smart contracts: sophisticated dataset curation and code transformation for artificial intelligence driven scanners

Blockchain is a database that stores data in a distributed manner, ensuring the integrity and immutability of data. Due to these features, blockchain technology has gained great popularity in the field of data storage and transfer and has brought many new technologies. One of these technologies is Ethereum, and one of its most important innovations has been smart contract technology. Smart contracts allow independent parties to transfer data securely by enabling them to transact in a decentralized and encrypted manner without the need for a third party. However, smart contracts can be exploited by malicious actors due to their vulnerabilities. Furthermore, once created, it is not possible to modify smart contracts due to the structure of the blockchain. Therefore, the detection of vulnerabilities is crucial. The main objective of this thesis is realized based on the Ethereum smart contracts verified in 2022. These smart contracts have been rigorously analyzed using the symbolic analysis method and their vulnerabilities have been identified. The obtained results play a critical role in the security of smart contracts. In this context, the obtained smart contract codes were subjected to transcoding and data preparation for the effective use of advanced artificial intelligence methods. Thanks to this novel approach, the dataset is transformed into a sequential structure, which allows AI tools to perform more in-depth analysis on smart contract security. The results of this study provide a solid foundation for future smart contract security research and inspire efforts to improve the security of smart contract technology.