DHGOrg

DiHypergraph: Analyzing and Visualizing GitHub Repositories

Dihypergraph offers a suite of tools designed to construct, visualize, and analyze data from GitHub repositories. Our primary data structure is a directed hypergraph, a generalization of a standard directed graph, which enables the representation of complex relationships in GitHub repositories. By using state-of-the-art techniques, we provide insightful visualizations and identify potential anomalies in commit messages using autoencoders.

Features

1. Construct Directed Hypergraphs from GitHub Data: Convert GitHub issues, commits, and pull requests into nodes of a dihypergraph, capturing rich relationships among them.
2. Visualize Dihypergraph Data: Plot the directed hypergraph using NetworkX and Matplotlib.
3. Anomaly Detection in Commit Messages: Utilizing the power of Transformer-based autoencoders to identify potentially anomalous commit messages.

Quick Start

# Clone the repository
git clone <repository_link>

# Navigate to the repository folder
cd dihypergraph

Usage

1. Constructing Dihypergraph

To generate a dihypergraph JSON representation from a GitHub repository:

from DHG import DHGConstructor

BASE_URL = "https://api.github.com/repos/YOUR_REPO_NAME/"
TOKEN = "YOUR_GITHUB_TOKEN"

my_dhg = DHGConstructor(BASE_URL, HEADERS, "output_name")
my_dhg.run_loop()

2. Visualize Dihypergraph

Visualize the dihypergraph and compute associations between issues:

from DHGVisualizer import DHGVisualizer

dhg_vis = DHGVisualizer("dihypergraph_output_name.json")
dhg_vis.create_graph()
dhg_vis.visualize()