WevN

WevN is a multi-tenant knowledge discovery and visualization platform that integrates semantic search, interactive graph visualization, and AI-assisted reasoning. It helps organizations explore and connect ideas across domains through an adaptive, RAG-based (Retrieval-Augmented Generation) architecture.

---

🚀 Features

• Semantic Search: Context-aware retrieval using embeddings from ChromaDB.
• Interactive Visualization: Real-time knowledge graph rendered with React Force Graph.
• LLM Integration: RAG pipeline with deepseek-r1:7b via Ollama for domain-specific responses.
• Multi-Tenant Design: Secure, isolated knowledge bases per organization.
• Node Automation: Automatic summarization and node creation from conversations.

---

🧱 Tech Stack

Frontend

• React 18+, Vite, TailwindCSS
• React Force Graph, React Markdown

Backend

• FastAPI, LangChain, ChromaDB, SQLite
• Sentence Transformers (all-mpnet-base-v2)
• Ollama for local LLM inference

Other

• JWT Authentication, Google OAuth
• Docker (optional)

---

⚙️ Setup

Prerequisites

• Node.js ≥ 16
• Python ≥ 3.8
• Ollama installed and model deepseek-r1:7b pulled

---

Frontend

git clone https://github.com/WevN-org/WevN.git
cd WevN/frontend
npm install
npm run dev

Backend

cd ../backend/server
python -m venv venv
source venv/bin/activate  # or .\venv\Scripts\activate on Windows
pip install -r requirements.txt
uvicorn server:app --host 0.0.0.0 --port 8000 --workers 2

📂 Structure

WevN/
├── backend/
│   └── server/
│       ├── main.py
│       ├── requirements.txt
│       └── ...
├── frontend/
│   ├── src/
│   └── ...
└── README.md