Time-aware Metapath Feature Augmentation for Ponzi Detection in Ethereum

TL;DR

This work addresses Ponzi detection in Ethereum by integrating temporal dynamics and heterogeneity through Time-aware Metapath Feature Augmentation (TMFAug). TMFAug merges symbiotic time-aware metapaths, refines and filters behavioral patterns, and aggregates their features into a homogeneous transaction graph to augment existing detectors without retraining. Experiments on Ethereum data show consistent performance gains across baselines and demonstrate the superiority of temporal information over timeless approaches. The approach highlights the value of incorporating heterogeneous temporal behavior patterns for blockchain fraud detection and points to future work on automatically learning metapaths to reduce manual engineering effort.

Abstract

With the development of Web 3.0 which emphasizes decentralization, blockchain technology ushers in its revolution and also brings numerous challenges, particularly in the field of cryptocurrency. Recently, a large number of criminal behaviors continuously emerge on blockchain, such as Ponzi schemes and phishing scams, which severely endanger decentralized finance. Existing graph-based abnormal behavior detection methods on blockchain usually focus on constructing homogeneous transaction graphs without distinguishing the heterogeneity of nodes and edges, resulting in partial loss of transaction pattern information. Although existing heterogeneous modeling methods can depict richer information through metapaths, the extracted metapaths generally neglect temporal dependencies between entities and do not reflect real behavior. In this paper, we introduce Time-aware Metapath Feature Augmentation (TMFAug) as a plug-and-play module to capture the real metapath-based transaction patterns during Ponzi scheme detection on Ethereum. The proposed module can be adaptively combined with existing graph-based Ponzi detection methods. Extensive experimental results show that our TMFAug can help existing Ponzi detection methods achieve significant performance improvements on the Ethereum dataset, indicating the effectiveness of heterogeneous temporal information for Ponzi scheme detection.

Figures (7)

• Figure 1: Schematic illustration of heterogeneous graphs, metapath, time-aware metapath, and symbiotic relationship.
• Figure 2: An example of a Ponzi scheme disguised as a game.
• Figure 3: Homogeneous and heterogeneous interaction graph.
• Figure 4: Interaction graphs of real Ponzi schemes in Ethereum.
• Figure 5: The overall framework of the Time-aware Metapath Feature Augmentation. The complete workflow proceeds as follows: (a) merging symbiotic relations to obtain the super metapaths; (b) refining the super metapaths based on the behavior, and filtering the super metapaths by Top-$K$; (c) aggregating the information along metapaths to the target nodes.

Theorems & Definitions (4)

• Definition 1: Heterogeneous Graph
• Definition 2: Metapath
• Definition 3: Time-aware Metapath
• Definition 4: Symbiotic Relationship