Graph embeddings for blockchain-induced networks

Abstract

In recent years, the rapid growth of blockchain technology has sparked massive curiosity and transformed various industries. Among the numerous blockchain platforms, Ethereum has gathered significant attention for its decentralized applications and smart contracts. Understanding Ethereum and its network interactions is a challenging task but with various methods at our disposal, such as graph embeddings, we gain valuable insight into its operations. Graph embeddings are powerful techniques in the realm of data representation, which have become a focal point in analyzing complex structures. By capturing the essence of graph’s structure and semantics, embeddings enable efficient analysis of vast networks. In the context of blockchain-induced networks, graph embeddings offer effective insights into the behavior and dynamics of transactions and addresses. In my Bachelor’s thesis, conducted under the guidance and support of Dr. Ferenc Beres and Marcell Nagy, I aim to explore the connection between graph embeddings and blockchain-induced networks from Ethereum with a binary classification problem on certain Ethereum accounts’ network interactions as graphs. Additionally, I analyze graph-level properties and employ dimensionality reduction techniques to visualize these networks in the embedding space.