Detecting Smart Contract Project Anomalies in Metaverse

Abstract

The metaverse virtual assets are carried by smart contract, thus detecting the potential smart contract vulnerabilities and malicious intension is critical to protect the metaverse digital assets and the investors’ confidence. However, existing research mainly focus on the Ethereum address features, which can hardly reveal the characteristics of the entire smart contract project. In this paper, we take the first step to detect smart contract anomalies on the granularity of smart contract project, and propose a method to identify the smart contract project’s subordinate addresses, and a model network which takes the runtime features of the smart contract project. We further apply our method on our dataset collected from projects deployed on Ethereum, and prove that our method could effectively recognize the similarity of smart contract projects, and identify the malicious transactions which trigger smart contract project anomalies.