Detailed analysis of Ethereum network on transaction behavior, community structure and link prediction

Area of Science:

• Blockchain Technology and Cryptocurrencies
• Network Analysis and Data Science
• Computational Social Science

Background:

• Ethereum, the second-largest cryptocurrency, possesses extensive transaction data but its network structure remains under-explored.
• Limited research exists on link predictability within the Ethereum transaction network.
• Understanding Ethereum's network dynamics is crucial for its future development and application.

Purpose of the Study:

• To introduce the Detailed Analysis of the Ethereum Network on Transaction Behavior, Community Structure, and Link Prediction (DANET) framework.
• To investigate wealth distribution, accumulation patterns, and community structure within the Ethereum Featured Transactional Network (EFTN).
• To evaluate the effectiveness of Variational Graph Auto-Encoders for link prediction on the EFTN.

Main Methods:

• Development and application of the DANET framework for comprehensive Ethereum network analysis.
• Exploration of wealth distribution and community detection algorithms on the EFTN.
• Implementation and testing of Variational Graph Auto-Encoders for link prediction tasks.

Main Results:

• The study provides insights into wealth distribution and community structures within the Ethereum network.
• Variational Graph Auto-Encoders achieved superior prediction accuracy for links on the EFTN.
• Network usage statistics were visualized, offering a basis for future development conjectures.

Conclusions:

• The DANET framework offers a robust methodology for analyzing complex blockchain networks like Ethereum.
• The findings highlight the potential of advanced machine learning models for understanding and predicting blockchain transaction behavior.
• The research provides valuable insights into the current utilization and future trajectory of the Ethereum network.