textgraph-lab

Upload any document and explore it as an interactive knowledge graph. Concepts are extracted via NLP co-occurrence analysis, clustered using the Louvain community detection algorithm, and ranked by betweenness centrality — then rendered as a force-directed graph you can explore and chat with.

What it does

1. Ingest — upload a PDF, image (OCR), URL (scraped), or plain text
2. Extract — tokenizes content, builds a co-occurrence graph of key concepts
3. Analyze — runs Louvain community detection (clusters) + betweenness centrality (importance ranking)
4. Visualize — renders an interactive Sigma.js force-directed graph: node size = importance, color = cluster
5. Chat — LLM-powered /insights and /chat endpoints let you query the graph in natural language

Stack

Layer	Tech
Frontend	Next.js 14, React, TypeScript, Sigma.js, Framer Motion, Tailwind CSS
Backend	Node.js, Express, TypeScript
Graph	graphology, graphology-communities-louvain, graphology-metrics
Ingestion	pdf-parse, Tesseract.js (OCR), Cheerio (URL scraping)
LLM	Anthropic Claude SDK, OpenAI SDK, LM Studio (local), Ollama (local)
Storage	SQLite (better-sqlite3)

Quick start

git clone https://github.com/And1zle/textgraph-lab
cd textgraph-lab

Backend:

cd backend
cp .env.example .env      # edit to set your LLM provider
npm install
npm run dev               # starts on http://localhost:3001

Frontend:

cd frontend
npm install
npm run dev               # starts on http://localhost:3000

Then open http://localhost:3000.

Configuration

Copy backend/.env.example to backend/.env and set your preferred LLM provider:

# Choose one: lmstudio | ollama | openai | anthropic
LLM_PROVIDER=lmstudio
LLM_MODEL=qwen/qwen3.5-9b

# LM Studio (local, free)
LMSTUDIO_BASE_URL=http://localhost:1234/v1
LMSTUDIO_API_KEY=lm-studio

# Ollama (local, free)
OLLAMA_BASE_URL=http://localhost:11434/v1
OLLAMA_MODEL=llama3.1:8b

# Cloud providers (require API keys)
OPENAI_API_KEY=
ANTHROPIC_API_KEY=

The tool works fully offline with LM Studio or Ollama — no API key required.

Project structure

textgraph-lab/
├── backend/
│   ├── src/
│   │   ├── index.ts          # Express server entry point
│   │   ├── routes/           # API route handlers
│   │   ├── services/         # LLM factory, graph engine, ingestion
│   │   ├── db/               # SQLite schema and queries
│   │   └── prompts/          # LLM system prompts
│   ├── .env.example
│   └── package.json
├── frontend/
│   ├── app/                  # Next.js app router pages
│   ├── components/
│   │   ├── GraphCanvas.tsx   # Sigma.js force-directed graph
│   │   ├── ChatPanel.tsx     # LLM chat interface
│   │   ├── InsightPanel.tsx  # Auto-generated insights
│   │   ├── TopicSidebar.tsx  # Community/cluster browser
│   │   └── AddSourcePanel.tsx# Document upload UI
│   └── package.json
└── start.bat                 # Windows: starts both servers

API endpoints

Method	Path	Description
POST	/api/analyses	Create a new analysis session
POST	/api/analyses/:id/ingest	Upload a document to an analysis
GET	/api/analyses/:id	Get graph data for an analysis
POST	/api/analyses/:id/insights	Generate LLM insights from the graph
POST	/api/analyses/:id/chat	Chat with the graph
POST	/api/analyses/:id/rebuild	Re-run graph extraction

License

MIT