EX-Graph: A Pioneering Dataset Bridging Ethereum and X

Authors: Qian Wang, Zhen Zhang, Zemin Liu, Shengliang Lu, Bingqiao Luo, Bingsheng He

ICLR 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Extensive experiments, including Ethereum link prediction, wash-trading Ethereum addresses detection, and X-Ethereum matching link prediction, emphasize the significant role of X data in enhancing Ethereum analysis.
Researcher Affiliation Academia Qian Wang, Zhen Zhang , Zemin Liu, Shengliang Lu, Bingqiao Luo, Bingsheng He National University of Singapore {qiansoc, zhen, zeminliu, lusl}@nus.edu.sg, luo.bingqiao@u.nus.edu, hebs@comp.nus.edu.sg
Pseudocode No The paper describes data collection, graph construction, and feature extraction processes, and refers to various algorithms and models, but it does not include any figure, block, or section labeled 'Pseudocode', 'Algorithm', or structured steps formatted like code for its own methodology.
Open Source Code Yes We have publicly released our dataset and codebase at https://github.com/Persdre/ EX-Graph, which includes the datasets, models, and settings necessary to reproduce our results.
Open Datasets Yes To access the data, please follow the links provided in the Git Hub README file and download the relevant files from Google Drive. The EX-Graph dataset is a collection of on-chain Ethereum transaction records and off-chain X following-follower data, arranged as graphs in DGL (Deep Graph Library) format.
Dataset Splits Yes We divide all Ethereum graph edges into training, validation, and test sets, following a 70/10/20 ratio based on edge timestamps.
Hardware Specification Yes All models in our experiments were implemented using Pytorch 2.0.0 in Python 3.9.16, and run on a robust Linux workstation. This system is equipped with two Intel(R) Xeon(R) Gold 6226R CPUs... The workstation is further complemented by a potent GPU setup, comprising eight NVIDIA Ge Force RTX 3090 GPUs, each providing 24.576 GB of memory.
Software Dependencies Yes All models in our experiments were implemented using Pytorch 2.0.0 in Python 3.9.16... To implement and train our graph-based models, we utilized the version 1.1.0 of the Deep Graph Library (DGL).
Experiment Setup Yes In this section, we present a list of the hyperparameters and final hyperparameters used in each baseline model for Ethereum link prediction. To ensure consistency across all models, we set the output node embedding size to 128 and the learning rate to 0.001.