ArXiv Research Paper Assistant

📌 Overview

An intelligent multi-agent application that automates academic research and literature review generation from arXiv papers. Built with AutoGen, OpenAI GPT-4o, and Streamlit.

🎯 Features

• Multi-Agent Collaboration: Two specialized agents working together (Research + Summarization)
• Real-time Streaming: Live output as agents process research
• Intelligent Search: Finds most relevant papers using arXiv API
• Professional Summaries: Generates formatted literature reviews
• User-Friendly Interface: Streamlit-based UI with configuration options
• Error Handling: Comprehensive logging and error management

🏗️ Project Structure

Arxiv_Research_Paper/
├── app.py                  # Main Streamlit application
├── pipeline.py             # Research orchestration
├── agents.py               # Agent initialization
├── constants.py            # Configuration constants
├── prompts.py              # Agent prompts & templates
├── utils.py                # Utility functions
├── requirements.txt        # Project dependencies
├── ARCHITECTURE.md         # Detailed architecture guide
└── .env                    # Environment variables (not in repo)

🚀 Quick Start

Prerequisites

• Python 3.8+
• OpenAI API key

Installation

1. Clone or download the project
2. Install dependencies:

pip install -r requirements.txt

3. Create a .env file in the project root:

OPENAI_API_KEY=your_openai_api_key_here

Running the Application

streamlit run app.py

The application will open in your browser at http://localhost:8501

📖 Usage

1. Select a Research Topic:

   • Choose from preset topics or enter a custom topic
   • Adjust maximum number of papers to retrieve

2. Start Research:

   • Click the "🚀 Start Research" button
   • Watch real-time results stream in

3. View Results:

   • ArXiv Research Agent fetches relevant papers
   • Summarizer Agent creates a literature review
   • Both responses are displayed in markdown format

🔧 Configuration

All configuration is centralized in constants.py:

• Model: GPT-4o (configurable)
• Max Results: 5 papers (adjustable via UI)
• Agent Names: ArxivResearchAgent, SummarizerAgent
• Max Turns: 2 (conversation rounds)

📊 Architecture

Agent Workflow

┌─────────────────────────────────────┐
│   User Input (Topic Selection)      │
└────────────┬────────────────────────┘
             │
             ▼
┌─────────────────────────────────────┐
│  ArxivResearchAgent                 │
│  - Formulates search query          │
│  - Fetches top relevant papers      │
│  - Returns JSON paper list          │
└────────────┬────────────────────────┘
             │
             ▼
┌─────────────────────────────────────┐
│  SummarizerAgent                    │
│  - Analyzes paper list              │
│  - Generates literature review      │
│  - Formats output in Markdown       │
└────────────┬────────────────────────┘
             │
             ▼
┌─────────────────────────────────────┐
│   Display Results (Streamlit UI)    │
└─────────────────────────────────────┘

🛠️ Core Modules

app.py

Main Streamlit interface with:

• Sidebar configuration panel
• Topic selection and input
• Real-time result display
• Error handling and user feedback

pipeline.py

Orchestration layer with:

• ResearchTeam class for agent management
• Async research execution
• Stream-based output handling

agents.py

Agent initialization with:

• OpenAI client setup
• ArXiv Research Agent
• Summarizer Agent
• Environment variable management

constants.py

Centralized configuration:

• Model settings
• Agent names
• UI configuration
• Default values

prompts.py

Prompt engineering:

• Research agent system message
• Summarizer agent system message
• Task templates

utils.py

Utility functions:

• arxiv_research(): Search arXiv API
• format_papers_for_display(): Format output
• Logging utilities

🔐 Environment Variables

Required in .env:

• OPENAI_API_KEY: Your OpenAI API key

📋 Dependencies

• autogen-agentchat: Multi-agent orchestration
• autogen-ext: AutoGen extensions
• autogen-ext[openai]: OpenAI integration
• streamlit: Web UI framework
• arxiv: arXiv API client
• python-dotenv: Environment variable management

See requirements.txt for specific versions.

⚠️ Important Notes

1. API Costs: Running this application will incur OpenAI API costs
2. Rate Limiting: Be mindful of arXiv API rate limits
3. API Key: Never commit .env file to version control
4. Async Execution: Application runs async tasks; ensure proper environment setup

🐛 Troubleshooting

"OPENAI_API_KEY not found"

• Ensure .env file exists in project root
• Verify the key format is correct
• Check file permissions

Streamlit Connection Issues

• Ensure all dependencies are installed
• Try running: pip install -r requirements.txt --upgrade
• Check if port 8501 is available

Slow Response Times

• Check internet connection
• Verify OpenAI API status
• Reduce max results for faster processing

📈 Performance

• Typical Research Time: 30-60 seconds for 5 papers
• Concurrent Operations: Supports async processing
• Memory Usage: Minimal for normal usage

🎓 Learning Resources

• AutoGen Documentation
• OpenAI API Reference
• arXiv API Documentation
• Streamlit Documentation

📝 License

This project is open source. Use and modify as needed.

🤝 Contributing

To improve this project:

1. Add tests (currently missing)
2. Implement result caching
3. Add data persistence
4. Improve error recovery
5. Enhance UI/UX

📞 Support

For issues or questions:

1. Check the ARCHITECTURE.md file for detailed information
2. Review error logs in the terminal
3. Verify environment setup

🌟 Future Enhancements

• Result caching and persistence
• User authentication and multi-user support
• Export research to PDF/Word
• Integration with reference management tools
• Advanced filtering and sorting options
• Custom prompt templates
• Batch research jobs