A Binary Classification Social Network Dataset for Graph Machine Learning

2025-03-04

AI Summary

The graph machine learning community lacks dedicated binary-classification benchmark datasets for social networks. To address this gap, we introduce BiSND—the first binary-classification social network benchmark dataset specifically designed for graph machine learning. BiSND is provided in both tabular and graph formats to support evaluation across conventional machine learning models (e.g., ensemble trees, XGBoost, MLP) and modern graph neural networks (e.g., GCN) as well as graph contrastive learning methods (e.g., BGRL, GRACE, DAENS). Its construction faithfully reflects real-world social structures and label distributions, enabling rigorous multi-paradigm model validation. Extensive experiments demonstrate consistent F1-scores of 67.66–70.15 across all evaluated models, confirming BiSND’s validity, generalizability, and robustness. This work fills a critical void in benchmark resources for binary social graph classification.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Data Mining & Knowledge Management: Graph Mining, Social Network Analysis & Community
• Reasoning under Uncertainty: Graphical Models

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Social Networks and Social Media: Fairness and bias in social network and social media analysis
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics

Abstract

Social networks have a vast range of applications with graphs. The available benchmark datasets are citation, co-occurrence, e-commerce networks, etc, with classes ranging from 3 to 15. However, there is no benchmark classification social network dataset for graph machine learning. This paper fills the gap and presents the Binary Classification Social Network Dataset ( extit{BiSND}), designed for graph machine learning applications to predict binary classes. We present the BiSND in extit{tabular and graph} formats to verify its robustness across classical and advanced machine learning. We employ a diverse set of classifiers, including four traditional machine learning algorithms (Decision Trees, K-Nearest Neighbour, Random Forest, XGBoost), one Deep Neural Network (multi-layer perceptrons), one Graph Neural Network (Graph Convolutional Network), and three state-of-the-art Graph Contrastive Learning methods (BGRL, GRACE, DAENS). Our findings reveal that BiSND is suitable for classification tasks, with F1-scores ranging from 67.66 to 70.15, indicating promising avenues for future enhancements.

Problem

• Lack of binary classification social network datasets for graph machine learning.
• Introduction of Binary Classification Social Network Dataset (BiSND).
• Evaluation of BiSND using diverse machine learning and graph neural network methods.

Innovation

• Introduces Binary Classification Social Network Dataset (BiSND)
• Utilizes diverse classifiers including Graph Neural Networks
• Employs state-of-the-art Graph Contrastive Learning methods