ResearchAI: AI-Powered Research Paper Assistant

ResearchAI Platform

📌 Overview

ResearchAI is an intelligent research assistant designed to help researchers save time when searching for and understanding academic papers. Instead of manually going through numerous papers, users can simply enter a query, and our AI agent will fetch and summarize the most relevant research papers.

🚀 How It Works

1. User enters a query → The system fetches around 30 relevant papers.
2. AI selects the best papers → Our AI agent filters and selects the top max papers papers using vector embeddings.
3. Summarization & Explanation → A summarizing AI agent generates concise summaries for each paper.
4. Interactive Chat → Users can engage in a chat-based environment to discuss the content of a selected paper.

🔥 Key Features

• AI-Powered Search: Finds and ranks the most relevant papers using AI techniques.
• Summarization Agent: Uses models to summarize papers efficiently.
• Interactive Chat Environment: Users can chat with an AI that understands the context of the selected research paper.
• Vector Search with FAISS: Papers are chunked and stored as vector embeddings for efficient retrieval.
• Chat Memory with MongoDB: Chat history is stored for context preservation.

🛠️ Tech Stack

Backend

• Flask: Serves as the backend framework.
• FAISS: Stores and retrieves vector embeddings.
• MongoDB: Stores chat history and memory.
• OAuth: Enables secure authentication.
• Node.js
• Express.js

Frontend

• Next.js: Used for server-side rendering and improved performance.
• React.js: Used for building an intuitive UI.
• Redux: Manages global state.
• Tailwind CSS: Styles the application.

📂 Project Setup

1. Install Dependencies

Backend

pip install -r requirements.txt

Frontend

pnpm install

2. Set Up Environment Variables

Rename example.env to .env and add necessary values (like GEMINI_API, MONGO_URL, PORT, etc.).

3. Run the Backend

For Flask:

python app.py

For Node.js/Express.js:

npm run dev

OR

node index.js

This starts the backend, which hosts the AI agents and APIs locally.

4. Run the Frontend

pnpm dev

This starts the frontend server, making the application accessible via the browser.

📺 Frontend Overview

• Landing Page: Entry point for users.
• Home Page: Displays recommended papers.
• Chat Environment: Allows users to interact with AI regarding specific papers.
• Summary Page: Displays AI-generated summaries.
• Analytics Dashboard: Provides insights and stats.

🎯 Challenges We Faced

• Handling Large Data: Processing large amounts of research data was challenging, so we split papers into chunks and stored them efficiently in a vector database (FAISS).
• Maintaining Chat Memory: To ensure the AI retains context, we stored chat histories in MongoDB.
• Efficient Paper Ranking: Developing a heuristic approach to rank the most relevant papers took effort.

🎥 Demo & Showcase

Link To The Backend

Link Of Ai Agents

Watch the project walkthrough