X-SPIDE: An eXplainable Machine Learning Pipeline for Detecting Smart Ponzi Contracts in Ethereum
Blockchain technology is revolutionizing digital asset exchange by eliminating the need for central authority control. However, the decentralized nature of blockchain attracts malicious actors, leading to the proliferation of financial scams, with Ponzi schemes being particularly prevalent. Conseque...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2025-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/11003052/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Blockchain technology is revolutionizing digital asset exchange by eliminating the need for central authority control. However, the decentralized nature of blockchain attracts malicious actors, leading to the proliferation of financial scams, with Ponzi schemes being particularly prevalent. Consequently, there is a growing need to develop automatic detection mechanisms for such scams. So far, the problem has been tackled by considering only classifier performances and with limited focus on the explanation and interpretation of the results. However, interpretability and explainability are crucial when classifier decisions may have economic consequences. This paper introduces X-SPIDE (XAI Smart Ponzi Identification and Detection), an explainable machine learning pipeline for Ponzi scheme detection within the Ethereum blockchain that aims to find the trade-off between performance and explainability. X-SPIDE allows for comparing the results of different classifiers, identifying a small set of features that offer strong performance, and understanding how these features contribute to classification — highlighting specific characteristics of malicious contracts. Moreover, we introduce and make publicly available a new comprehensive dataset comprising 7446 smart contracts, incorporating features derived from transaction history, creation, and deployment bytecodes to train and test our pipeline. |
|---|---|
| ISSN: | 2169-3536 |