Scammer identification using CatBoost in smart contract for enhancing security in blockchain network.

In the rapidly upgrading world, blockchain is evolving where smart contract emerges as an important aspect that is automated to execute when the pre-conditions are satisfied. Properties like self-execution, non-reversible and non-terminating properties are mainly developed to eliminate the presence of third parties. However, such properties are misused by scammers to create malicious contracts, causing financial losses. Self-destruction property is also misused whereas the scammers remain anonymous after causing the destruction. So, it is necessary to identify the fraudulent nodes before they escape acquiring large sum of money. In this paper, the smart contract Ethereum Fraud Detection dataset was tested against various machine learning algorithms. Categorial Boosting (CatBoost) emerges as the best algorithm having a balanced tree structure contributing to its superior performance. Finally, comparative analysis shows that CatBoost outperforms obtaining 96.85% accuracy and obtains 97% when normalizing continuous features. Hence, this research addresses a way to detect scammers and contributes to interacting with safer smart contracts in blockchain networks. [ABSTRACT FROM AUTHOR]