一句话：提出 DeepTx，通过模拟待确认交易、提取多模态特征并借助多 LLM 共识推理，在用户确认前实时检测 Web3 钓鱼交易。
Web3 生态中的钓鱼攻击日益复杂，攻击者利用欺骗性的合约逻辑、恶意前端脚本以及代币授权模式，在用户确认交易前难以被察觉。本文提出 DeepTx，一个实时交易分析系统：它先模拟待处理的交易，从中提取行为特征、上下文特征与 UI 特征，再调用多个大语言模型（LLM）对交易意图进行推理；为保证判断的稳健性与可解释性，系统采用带自我反思（self-reflection）的共识机制来融合多个模型的结论。作者在自建的钓鱼数据集上对 DeepTx 进行评估，实验表明其能够同时取得较高的精确率与召回率，并提供了演示视频。
Phishing attacks in Web3 ecosystems are increasingly sophisticated, exploiting deceptive contract logic, malicious frontend scripts, and token approval patterns. We present DeepTx, a real-time transaction analysis system that detects such threats before user confirmation. DeepTx simulates pending transactions, extracts behavior, context, and UI features, and uses multiple large language models (LLMs) to reason about transaction intent. A consensus mechanism with self-reflection ensures robust and explainable decisions. Evaluated on our phishing dataset, DeepTx achieves high precision and recall (demo video: https://youtu.be/4OfK9KCEXUM).
DOI 原文 ·
@inproceedings{paperbot9197,
title = {DeepTx: Real-Time Transaction Risk Analysis via Multi-Modal Features and LLM Reasoning},
author = {Yixuan Liu and Xinlei Li and Yi Li},
booktitle = {ASE},
year = {2025},
doi = {10.1109/ase63991.2025.00366}
}