Blockchain and machine learning for fraud detection in e-commerce transactions

The research suggests employing algorithms based on machine learning (ML) in conjunction with blockchain technology and smart contracts for detecting fraud in e-commerce. Blockchain technology is used to simplify the process and protect the privacy of information. Cooperative data from blockchain-connected enterprises is used to gradually improve a ML framework. According to the degree of complexity in modifying the model, a reward system is provided to incentivize companies. An effective mining criteria for blocks is suggested. To evaluate its effectiveness, the blockchain system is put through a series of tests with varying degrees of complexity and data quantities. The model’s total Fbeta score was 98.22% and its evaluation correctness was 98.93% across eight progressive upgrades. The study demonstrates that mining time is influenced by both the amount of information collected as well as the level of blockchain complexity. There is a positive association between mining time and challenge level for ranges of difficulty below five. A block in the network may be mined in less than a second for challenging levels 2 and 3.