GitHub Stargazers | Building Graph- and Edge-level Prediction Algorithms for Developer Social Networks

TL;DR

This paper utilizes graph neural networks to analyze GitHub developer networks, effectively classifying communities and predicting potential collaborations to enhance understanding of open-source social structures.

Contribution

It introduces a novel approach combining community segmentation and edge-level prediction algorithms tailored for developer social networks.

Findings

Accurate community segmentation based on repository focus

Effective prediction of new developer connections

Improved understanding of developer collaboration patterns

Abstract

Analyzing social networks formed by developers provides valuable insights for market segmentation, trend analysis, and community engagement. In this study, we explore the GitHub Stargazers dataset to classify developer communities and predict potential collaborations using graph neural networks (GNNs). By modeling 12,725 developer networks, we segment communities based on their focus on web development or machine learning repositories, leveraging graph attributes and node embeddings. Furthermore, we propose an edge-level recommendation algorithm that predicts new connections between developers using similarity measures. Our experimental results demonstrate the effectiveness of our approach in accurately segmenting communities and improving connection predictions, offering valuable insights for understanding open-source developer networks.