Apppublic

Sudhanshu1624/Layer10-KG

sourceHugging Faceupdated 7mo agoView on Hugging Face

0likes

App README

Layer10 · Organizational Knowledge Graph

A modern, full-stack system designed to capture, deduplicate, and visualize organizational memory (entities, decisions, and relationships) from GitHub project history.

🚀 Quick Start

1. Prerequisites

—Python 3.10+ (for backend & pipeline)
—Node.js 18+ & npm (for frontend)
—Gemini API Key (for running the extraction pipeline)

2. Environment Setup

Create a .env file in the root directory:

env

GITHUB_TOKEN=your_github_token
GEMINI_API_KEY=your_gemini_api_key

3. Run the Backend (FastAPI)

powershell

cd backend
pip install -r requirements.txt
python -m uvicorn main:app --reload --port 8000

API available at: http://localhost:8000

4. Run the Frontend (React)

powershell

cd frontend
npm install
npm start

UI available at: http://localhost:3000

🏗️ Architecture Overview

The system is split into four distinct phases, migrating from raw GitHub data to an interactive knowledge graph.

1. Ingestion (`pipeline/ingest.py`)

Fetches closed GitHub issues and comments into a unified data/corpus.json.

2. Extraction (`pipeline/extract.py` + `pipeline/ontology.py`)

Uses structured semantic analysis to extract entities (Person, Decision, Topic, Artifact) and relationships with verbatim textual evidence.

3. Graph Construction (`pipeline/build_graph.py`)

Builds a directed NetworkX graph. Implements ID Normalization to deduplicate entities across batches (e.g., merging "@john_doe" and "John Doe").

4. Visualization (FastAPI + React)

—Backend API: A FastAPI server that performs graph filtering, metrics calculation, and serves PyVis HTML.
—Frontend UI: A professional React dashboard (dark theme) featuring a live Knowledge Graph, interactive Analytics charts, and a Node Explorer for evidence traversal.

📄 Ontology & Relationship Contract

Entity Type	Description
`Person`	Team members or contributors mentioned in discussions.
`Topic`	Abstract concepts, requirements, or design patterns.
`Artifact`	Code artifacts, URLs, documentation, or specific versions.
`Decision`	Resolved choices, rejections, or terminal states of a discussion.

Every relationship in the graph is grounded. In the Node Explorer, you can view the exact quote and source URL that justifies every edge, ensuring verifiable data lineage.

🛠️ Project Structure

Layer10/
├── 📂 backend/           # FastAPI server & graph service logic
├── 📂 frontend/          # React UI (CRA) & API integration layer
├── 📂 pipeline/          # Data generation & extraction scripts
├── 📂 data/              # Persistent stores (corpus, extractions, final graph)
└── README.md             # This document