ArXivMind 📚

ArXivMind is an intelligent, multi-agent system designed to accelerate academic research. Built with Microsoft's AutoGen framework, it automates the process of finding and summarizing scholarly articles from arXiv. The system deploys two specialized AI agents that collaborate to conduct a literature review on any given topic, delivering a concise and structured report directly to the user.

This project offers two modes of operation:

1. A command-line interface (backend.py) for running a predefined search query.
2. An interactive web application (streamlit.py) built with Streamlit for a dynamic, user-friendly experience.

---

✨ Key Features

• Multi-Agent Collaboration: Utilizes a search_agent to craft optimal queries and retrieve papers from arXiv, and a summarizer agent to analyze the findings and generate a high-quality literature review.
• Automated Literature Review: Generates a markdown-formatted report that includes an introduction, detailed summaries of each paper (with links, authors, contributions, and future scope), and a final takeaway.
• Interactive Web UI: Features a clean and simple interface built with Streamlit, allowing users to input their research topic and select the number of papers to review.
• Powered by OpenAI: Leverages the gpt-4o-mini model for intelligent query generation and sophisticated text summarization.
• Real-time Streaming: Streams the agents' conversation and final report in real-time, providing transparency into the research process.

---

🎬 Demo

See ArXivMind in action! The demo shows how to use the interactive Streamlit web application to generate a literature review. Click the image below to watch the video.

---

🛠️ Getting Started

Follow these instructions to set up and run ArXivMind on your local machine.

Prerequisites

• Anaconda or Miniconda installed.
• An OpenAI API key.

1. Clone the Repository

First, clone the project repository from GitHub.

git clone [https://github.com/YOUR_GITHUB_USERNAME/ArXivMind.git](https://github.com/YOUR_GITHUB_USERNAME/ArXivMind.git)
cd ArXivMind

Note: Remember to replace YOUR_GITHUB_USERNAME with your actual GitHub username.

2. Create and Activate Conda Environment

Create a dedicated Conda environment for the project using Python 3.12.

conda create -n arxivmind python=3.12 -y
conda activate arxivmind

3. Install Dependencies

Install all the required Python packages from the requirements.txt file.

pip install -r requirements.txt

4. Set Up Environment Variables

You need to provide your OpenAI API key for the agents to function.

1. Create a file named .env in the root directory of the project.
2. Add your OpenAI API key to the file as shown below:

OPENAI_API_KEY="sk-..."

---

🚀 Usage

ArXivMind can be run in two different modes.

1. Terminal Mode (backend.py)

This mode runs a hardcoded literature review task directly in your terminal. It's useful for testing the backend logic. By default, it searches for 5 papers on "Artificial Intelligence".

To run it, execute the backend.py script:

python backend.py

The script will stream the conversation between the search_agent and the summarizer, followed by the final literature review.

2. Interactive Web App (streamlit.py)

This is the primary mode for using ArXivMind. It launches a local web server with a user-friendly Streamlit interface.

To start the web application, run the following command in your terminal:

streamlit run streamlit.py

This will open a new tab in your web browser at http://localhost:8501. You can then enter a research topic, select the number of papers, and click "Search" to begin.

---

📂 Project Structure

.
├── backend.py            # Script for terminal-based agent interaction
├── streamlit.py          # Script to launch the interactive Streamlit web UI
├── requirements.txt      # Python dependencies
├── .env                  # File for API keys (you need to create this)
└── README.md             # This file