A Retrieval-Augmented Generation (RAG) system for fetching, indexing, and summarizing news articles using local AI models.

RAG News Summarizer demonstrates a complete RAG pipeline — from data ingestion to semantic search to LLM-powered summarization — all running locally on your machine.

This project was intentionally built as a simple, educational implementation of RAG concepts.

Modern AI applications increasingly rely on RAG architectures, but many tutorials are either:

• Too abstract (theory without working code)
• Too complex (production systems with overwhelming features)
• Cloud-dependent (requiring API keys and paid services)

RAG News Summarizer fills the gap by providing:

• ✅ A complete, working implementation you can run locally
• ✅ Clean, readable code that's easy to understand and modify
• ✅ No API keys required — uses local models only
• ✅ A foundation to build upon for your own projects

This is ideal for learning, portfolio projects, or as a starting point for production systems.

• 🔍 Semantic Search — Find relevant articles using vector similarity
• 📰 Multi-Source Ingestion — Fetches from BBC, Reuters, TechCrunch, and more
• 🤖 Local LLM Integration — Runs entirely on your machine with Ollama
• 💾 Persistent Storage — ChromaDB vector database for efficient retrieval
• 🎨 Modern Web UI — Beautiful Streamlit interface with real-time updates
• 🔒 Privacy-First — No data leaves your machine

┌─────────────────────────────────────────────────────────────────┐
│                     Streamlit Web UI                            │
└─────────────────────────────────────────────────────────────────┘
                              │
                              ▼
┌─────────────────────────────────────────────────────────────────┐
│                   RAG Pipeline (LangChain)                      │
│  ┌───────────────┐  ┌────────────────┐  ┌────────────────────┐  │
│  │ News Fetcher  │  │ Vector Store   │  │ LLM Chain          │  │
│  │ (RSS Parser)  │  │ (ChromaDB)     │  │ (Ollama)           │  │
│  └───────────────┘  └────────────────┘  └────────────────────┘  │
└─────────────────────────────────────────────────────────────────┘
         │                    │                    │
         ▼                    ▼                    ▼
   ┌──────────┐      ┌──────────────┐      ┌─────────────┐
   │ RSS Feeds│      │ Sentence     │      │ Ollama      │
   │ (Free)   │      │ Transformers │      │ (Llama 3.2) │
   └──────────┘      └──────────────┘      └─────────────┘

For detailed architecture documentation, see docs/ARCHITECTURE.md.

• Python 3.10 or higher
• Ollama installed

# 1. Clone the repository
git clone https://github.com/charanpool/rag-news-summarizer.git
cd rag-news-summarizer

# 2. Create virtual environment
python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

# 3. Install dependencies
pip install -r requirements.txt

# 4. Set up Ollama (in a separate terminal)
ollama pull llama3.2
ollama serve

# 5. Run the application
streamlit run app/main.py

Open your browser at http://localhost:8501 🎉

1. Click "🔄 Fetch & Index News" in the sidebar
2. Wait for articles to be fetched and indexed
3. Ask a question like "What's the latest in AI?"
4. View the AI-generated summary with source citations

RAG_News_Summarizer/
├── app/                    # Application code
│   ├── __init__.py
│   ├── config.py           # Configuration settings
│   ├── embeddings.py       # Embedding generation
│   ├── main.py             # Streamlit web interface
│   ├── news_fetcher.py     # RSS feed parser
│   ├── rag_chain.py        # RAG pipeline logic
│   └── vector_store.py     # ChromaDB operations
├── data/                   # Runtime data (gitignored)
│   └── chroma_db/          # Vector database storage
├── docs/                   # Documentation
│   ├── ARCHITECTURE.md     # Technical architecture
│   └── CONCEPTS.md         # Core RAG concepts
├── tests/                  # Test suite
│   └── test_rag.py
├── .env.example            # Environment template
├── .gitignore
├── CONTRIBUTING.md         # Contribution guidelines
├── LICENSE                 # MIT License
├── README.md               # This file
├── ROADMAP.md              # Future enhancement plans
└── requirements.txt        # Python dependencies

RSS Feeds → Parse Articles → Chunk Text → Generate Embeddings → Store in ChromaDB

User Query → Embed Query → Semantic Search → Retrieve Top-K Articles

Retrieved Context + User Query → LLM Prompt → Generated Summary → Display

For a deeper dive, see docs/CONCEPTS.md.

Key settings in app/config.py:

Override via .env file:

cp .env.example .env
# Edit .env with your preferences

Add custom sources in app/config.py:

RSS_FEEDS = {
    "Your Source": "https://example.com/rss/feed.xml",
}

# Run all tests
pytest tests/ -v

# Run with coverage
pytest tests/ --cov=app --cov-report=html

This project follows intentionally lean design principles:

This project skips a few things on purpose to stay beginner-friendly:

• User authentication
• Cloud deployments
• Complex caching layers
• Microservices architecture
• Kubernetes configurations

Want to add these? Check out ROADMAP.md for ideas!

# Start Ollama server
ollama serve

# Verify it's running
curl http://localhost:11434/api/tags

# Pull the required model
ollama pull llama3.2

• Use a smaller embedding model in config
• Reduce CHUNK_SIZE
• Use a smaller Ollama model: llama3.2:1b

Ensure you're running from the project root:

cd /path/to/RAG_News_Summarizer
source venv/bin/activate
streamlit run app/main.py

Contributions are welcome! Please see CONTRIBUTING.md for guidelines.

Areas where help is appreciated:

• Improving heuristic parsing
• Adding test fixtures
• Documentation improvements
• UI/UX enhancements

See ROADMAP.md for planned features including:

• Hybrid search (semantic + keyword)
• Multiple LLM provider support
• REST API endpoints
• Docker containerization
• Enhanced UI features

This project is licensed under the MIT License — see the LICENSE file for details.

• ✅ Free for personal and commercial use
• ✅ Modify and distribute freely
• ✅ Attribution required

• LangChain — RAG framework
• Ollama — Local LLM inference
• ChromaDB — Vector storage
• Sentence-Transformers — Embeddings
• Streamlit — Web interface

If you find this project useful:

• ⭐ Star the repository
• 🐛 Report issues you encounter
• 💡 Share ideas for improvements
• 🧑‍💻 Contribute code or documentation

Built with ❤️ for learning and demonstration purposes