IGNITE Toolkit Hub

A Unified Promotional Web Hub for PyOD, PyGDebias, and PyGIP

An interactive project page that frames three trustworthy-ML libraries as one coherent toolkit ecosystem for anomaly detection, fair graph learning, and graph model IP protection.

Open Webpage · View HTML · Why This Hub · Toolkit Story · Key Features · Quick Start · Publish · Metrics

PyOD · PyGDebias · PyGIP in one adoption-focused, interactive narrative

---

Why This Hub

The goal of this page is to turn three separate library introductions into a single promotional story: a trustworthy machine-learning toolkit ecosystem with measurable community adoption and practical onboarding.

Instead of writing a separate user-study section, the page emphasizes visible, quantifiable signals: GitHub stars, forks, PyPI versions, public download badges, algorithm coverage, dataset coverage, and quick-start commands. This makes the impact immediately legible to reviewers, visitors, and potential users.

Aim	How the page addresses it
Backend / unified architecture	Presents PyOD, PyGDebias, and PyGIP as services behind a shared metadata registry and API layer
Frontend / interface	Provides a polished responsive page, toolkit comparison cards, a pipeline view, install tabs, and a lightweight Q&A assistant
Evaluation / measurable impact	Replaces user-study prose with adoption metrics, badge-driven popularity signals, and project-scale facts

---

Toolkit Story

Toolkit	Role in the unified ecosystem	Promotional signal
PyOD	Outlier and anomaly detection module	60+ detectors, large public download footprint, strong GitHub adoption
PyGDebias	Fair graph learning and graph debiasing module	26 graph datasets and 13 fair graph mining algorithms
PyGIP	Graph model IP protection module	Modular graph model extraction attack and defense workflows

Together, they support a clean narrative:

Detect anomalous behavior -> audit graph fairness -> protect graph models from extraction risk

---

Key Features

• Badge-first adoption story: Live Shields and PePy badges show stars, forks, PyPI versions, and public download signals directly on the page.
• Unified product framing: The three libraries are introduced under one coherent trustworthy-ML platform narrative.
• Interactive Q&A panel: Visitors can ask common questions about installation, toolkit selection, graph fairness, anomaly detection, and model extraction risk.
• Copyable quick starts: Each toolkit has an install-oriented code panel designed for fast onboarding.
• Responsive promotional design: The page is built as a static site that works across desktop and mobile without a build step.
• GitHub Pages-ready: The folder can be published as a static site directly.

Adoption-first cards Each toolkit card combines positioning text with live popularity badges so visitors immediately see community traction.	Unified platform pipeline The architecture section explains how independent packages become one registry, API layer, visual builder, and assistant experience.
Question answering demo The lightweight assistant answers practical questions such as which library to use, how to install it, and how to compare them.	No user-study burden The page foregrounds quantifiable metrics and public adoption signals, matching the requested promotional direction.

---

Quick Start

Open the site locally:

cd ignite-toolkits-site
python3 -m http.server 4173

Then visit:

http://localhost:4173

You can also open index.html directly in a browser. The local server is recommended because it mirrors how the page will behave after publishing.

---

Toolkit Install Commands

pip install pyod
pip install pygdebias
pip install PyGIP

Minimal examples shown on the page:

from pyod.models.iforest import IForest

clf = IForest()
clf.fit(X_train)
scores = clf.decision_scores_

from pygdebias.datasets import Bail

dataset = Bail()
adj, features, labels, idx_train, idx_val, idx_test, sens = dataset.load_data()

---

Project Structure

ignite-toolkits-site/
  index.html
  styles.css
  app.js
  README.md

File	Purpose
index.html	Page structure, toolkit cards, badge embeds, install panels
styles.css	Responsive promotional visual design
app.js	Q&A presets, tab switching, copy-to-clipboard behavior
README.md	GitHub-facing project overview and publishing notes

---

Publish The Page

To publish this page with GitHub Pages:

Settings -> Pages -> Deploy from a branch
Branch: main
Folder: /ignite-toolkits-site

If GitHub Pages only allows /docs or repository root in your setup, move the three site files into docs/ and choose:

Settings -> Pages -> Deploy from a branch
Branch: main
Folder: /docs

---

Source Projects

Project	GitHub	Package
PyOD	yzhao062/pyod	pyod on PyPI
PyGDebias	yushundong/PyGDebias	pygdebias on PyPI
PyGIP	LabRAI/PyGIP	PyGIP on PyPI

---

Repository Metrics

PyOD

PyGDebias

PyGIP

---

IGNITE Toolkit Hub makes adoption, usability, and the unified trustworthy-ML story visible in one place.