GitHunt
SH

shakilahmedemon/rag-research-assistant

The Advanced RAG Academic Assistant is a sophisticated Retrieval-Augmented Generation (RAG) application designed for researchers and students. It allows users to upload multiple academic PDF papers, which are then processed, chunked, and embedded into a local vector store.

academic-researchacademic-research-toolsgoogle-geminigoogle-gemini-apiinformation-retrievalnlpnlp-machine-learningpdf-analysispythonpython3ragrag-chatbotrag-pipelinesemantic-searchsemantic-segmentationstreamlitstreamlit-applicationstreamlit-dashboardstreamlit-webappvector-embeddings

Advanced RAG Academic Assistant

An intelligent, professional-grade Research Paper Analysis & Question Answering system. This tool utilizes a Retrieval-Augmented Generation (RAG) pipeline to help users synthesize information from complex academic documents.

Key Features

  1. Multi-PDF Processing: Upload and index multiple research papers simultaneously.
  2. Semantic Retrieval: Uses all-MiniLM-L6-v2 embeddings to find the most relevant context for your questions, not just keyword matches.
  3. Gemini-Powered Synthesis: Leverages Google Gemini (Pro) to generate structured, academic-style reports (Introduction, Methodology, Results, Discussion, Conclusion).
  4. Automatic Citations: Every generated answer includes inline citations (e.g., [paper.pdf, Chunk 1]) to prevent hallucination.
  5. PDF Export: Download your generated research report as a professionally formatted PDF.
  6. Modern UI: Built with a custom-styled Streamlit interface featuring CSS animations and responsive layouts.

Technical Architecture

  1. Ingestion: pypdf extracts text from uploaded PDFs.
  2. Chunking: Documents are split into 1000-character segments with a 200-character overlap to preserve context.
  3. Embeddings: sentence-transformers creates vector representations of text chunks stored in numpy arrays.
  4. Retrieval: Performs Cosine Similarity search to find the top $k$ relevant chunks.
  5. Generation: Chunks are passed to the gemini-pro model with strict "Anti-Hallucination" prompting.

Getting Started

1. Prerequisites

  1. Python 3.9 or higher.
  2. A Google AI Studio API Key.

2. Installation

Clone the repository and install the dependencies:

3. Configuration

The system uses a config.py file for global settings. You can adjust:

  1. CHUNK_SIZE: Default 1000.
  2. TOP_K: Number of document chunks to retrieve (Default 5).
  3. LLM_MODEL: Default "gemini-pro".

4. Running the App

  • streamlit run app.py

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

Author

AHMED MD SHAKIL

Studying Master's in Software Engineering at Yangzhou University, China

Contributors

SH
MIT License
Created January 12, 2026
Updated January 15, 2026