Fraud detection in DeFi using transaction graphs and social sentiment analysis with a Web ۳ and machine learning approach

In today’s world, with the increasing growth of Web۳ and decentralized finance (DeFi) technologies, financial fraud has become one of the main challenges in this ecosystem. Traditional detection systems often rely solely on the transactional or structural features of a single blockchain, which limits the detection of frauds committed across chains. In this paper, we introduce an intelligent framework for detecting fraud in Web۳ and decentralized finance systems using transaction graphs and social sentiment analysis in social media. In this method, we connect to data through blockchain Application Programming Interface (APIs) such as Etherscan and Moralis and analyze transactional and textual data simultaneously using machine learning algorithms. In particular, this framework identifies abnormal patterns in transactions and user behavior and can effectively detect fraud and suspicious activities. Research results show that combining graph analysis and sentiment analysis can significantly increase the accuracy and speed of detection compared to existing methods and play an important role in Web۳ security.

کلیدواژه ها:

Decentralized financial platforms ، fraud detection ، graph neural networks ، sentiment analysis ، blockchain security ، machine learning ، cross-chain ، token fraud