AI Research Paper Insight & Gap Identifier – RAG, Sentence Transformers, FastAPI & Gradio

A web application that uses Retrieval-Augmented Generation (RAG) to analyze research papers and extract key insights or identify research gaps.

Features

• PDF Analysis: Upload any research paper in PDF format
• Key Insights Extraction: Automatically identify and summarize the main findings
• Research Gaps Detection: Highlight limitations and areas for future research
• User-Friendly Interface: Simple web UI powered by Gradio
• API Access: FastAPI backend allows programmatic access

Technology Stack

• FastAPI: Modern, high-performance web framework
• Gradio: Simple UI for machine learning models
• LangChain: Framework for LLM applications
• Azure OpenAI: Powerful language model integration
• FAISS: Vector similarity search for document retrieval
• HuggingFace Embeddings: Sentence transformers for text representation

Getting Started

Prerequisites

• Python 3.9+
• Azure OpenAI API access

Installation

1. Clone the repository:

   git clone https://github.com/your-username/AI-Research-Analyzer.git
   cd AI-Research-Analyzer

2. Install dependencies:

   pip install -r requirements.txt

3. Set up environment variables:

   cp .env .env
   # Edit .env with your Azure OpenAI API key and endpoint

Running the Application

Run the application with:

python -m app.main

The application will be available at http://127.0.0.1:8000

Usage

1. Access the web interface at http://127.0.0.1:8000
2. Upload a research paper in PDF format
3. Select analysis type: "Key Insights" or "Research Gaps"
4. View the generated analysis

How It Works

The application uses a Retrieval-Augmented Generation (RAG) architecture:

1. Document Processing: PDFs are loaded and split into manageable chunks
2. Vector Embedding: Text chunks are converted to vector embeddings
3. Retrieval: When a query is made, relevant chunks are retrieved
4. Generation: Retrieved content is sent to the LLM with a specialized prompt
5. Response: A structured analysis is returned to the user

Future Improvements

• Support for additional document formats (DOCX, TXT)
• Batch processing of multiple papers
• More analysis types (methodology critique, literature comparison)
• Visualization of document relationships and key concepts

License

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

Acknowledgments

• Built with LangChain
• Powered by Azure OpenAI models
• Interface created with Gradio