Abstract

The rampant fraudulent activities on Ethereum hinder the healthy developmentof the blockchain ecosystem, necessitating the reinforcement of regulations.However, multiple imbalances involving account interaction frequencies andinteraction types in the Ethereum transaction environment pose significantchallenges to data mining-based fraud detection research. To address this, wefirst propose the concept of meta-interactions to refine interaction behaviorsin Ethereum, and based on this, we present a dual self-supervision enhancedEthereum fraud detection framework, named Meta-IFD. This framework initiallyintroduces a generative self-supervision mechanism to augment the interactionfeatures of accounts, followed by a contrastive self-supervision mechanism todifferentiate various behavior patterns, and ultimately characterizes thebehavioral representations of accounts and mines potential fraud risks throughmulti-view interaction feature learning. Extensive experiments on real Ethereumdatasets demonstrate the effectiveness and superiority of our framework indetecting common Ethereum fraud behaviors such as Ponzi schemes and phishingscams. Additionally, the generative module can effectively alleviate theinteraction distribution imbalance in Ethereum data, while the contrastivemodule significantly enhances the framework's ability to distinguish differentbehavior patterns. The source code will be available inhttps://github.com/GISec-Team/Meta-IFD.