Scholarlink

<p>Fraudulent activity on Ethereum, the second-largest blockchain platform, presents escalating risks to financial systems operating in decentralized environments. Despite Ethereum’s wide adoption, research dedicated specifically to fraud detection within its network remains limited. This systematic review analyzes recent machine learning (ML) and deep learning (DL) approaches for detecting Ethereum-based fraud. A structured search of IEEE Xplore and ACM Digital Library from 2017 to 2024 yielded 305 studies, of which 58 met strict inclusion criteria. The results show a strong reliance on the Kaggle Ethereum Fraud Detection Dataset, which limits the external validity of many findings. XGBoost, Random Forest (RF), and Support Vector Machine (SVM) are the most frequently applied algorithms, with XGBoost consistently delivering top performance in classification tasks. However, few studies address real-time detection, and only a small number develop models specifically designed for Ethereum’s transaction structure. Compared to Bitcoin, Ethereum fraud detection remains underexplored. This review not only synthesizes current techniques and datasets but also identifies concrete research gaps, including the need for real-time, on-chain detection methods, diversified and representative datasets, and graph-based models that capture the structural properties of Ethereum’s network.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Blockchain-based fraud detection: A systematic review of Ethereum network applications

  • 期刊论文

  • 发表日期：

    2025年11月3日

查看全文
  • Rawan Ghnemat,
  • Hatem Mosa

摘要

Fraudulent activity on Ethereum, the second-largest blockchain platform, presents escalating risks to financial systems operating in decentralized environments. Despite Ethereum’s wide adoption, research dedicated specifically to fraud detection within its network remains limited. This systematic review analyzes recent machine learning (ML) and deep learning (DL) approaches for detecting Ethereum-based fraud. A structured search of IEEE Xplore and ACM Digital Library from 2017 to 2024 yielded 305 studies, of which 58 met strict inclusion criteria. The results show a strong reliance on the Kaggle Ethereum Fraud Detection Dataset, which limits the external validity of many findings. XGBoost, Random Forest (RF), and Support Vector Machine (SVM) are the most frequently applied algorithms, with XGBoost consistently delivering top performance in classification tasks. However, few studies address real-time detection, and only a small number develop models specifically designed for Ethereum’s transaction structure. Compared to Bitcoin, Ethereum fraud detection remains underexplored. This review not only synthesizes current techniques and datasets but also identifies concrete research gaps, including the need for real-time, on-chain detection methods, diversified and representative datasets, and graph-based models that capture the structural properties of Ethereum’s network.