GitHub Stargazers: Graph-Based Analysis of Developer Networks

Overview

This project explores social networks of developers using the GitHub Stargazers dataset from the Stanford Network Analysis Project (SNAP). The dataset consists of 12,725 graphs representing developer interactions based on repository stars. This analysis focuses on:

• Graph classification: Identifying whether a developer network is associated with web development or machine learning repositories.
• Recommendation system: Predicting potential new connections among developers using Graph Neural Networks (GNNs).

Project Structure

• recommendation_system.ipynb - Implements a GraphSAGE-based recommendation model to predict new developer connections.
• repo_classification.ipynb - Trains a Graph Convolutional Network (GCN) to classify networks into web development or machine learning communities.
• requirements.txt - Contains all necessary dependencies to run the notebooks.

Installation

To run this project, clone the repository and install the dependencies:

git clone https://github.com/karishmathakrar/stargazers.git
cd stargazers
pip install -r requirements.txt

Running the Notebooks

Ensure you have Jupyter Notebook installed and launch it:

jupyter notebook

Then, open and excute:

• recommendation_system.ipynb for edge prediction.
• repo_classification.ipynb for classification of developer networks.

License

This project is released under the MIT License.