Hybrid RAG Approach with Knowledge Graph

Overview

This project implements a Hybrid Retrieval-Augmented Generation (RAG) approach with knowledge graphs for interacting with PDF documents. The system integrates several advanced techniques to enhance interaction with document-based data, combining text extraction, semantic understanding, and generative response generation.

Flowchart

Below is the flowchart representing the process:

File Structure

The project consists of the following Python scripts:

1. extraction.py

• Purpose: Extracts text from PDF documents.
• Dependencies: Uses the llama-Index library for parsing and extracting text.
• How It Works:
  • Converts PDF content into a machine-readable format.
  • Cleans and structures the extracted text for further processing.

2. graphcreation.py

• Purpose: Creates and populates a knowledge graph with entities and relationships.
• Dependencies: Requires libraries for knowledge graph creation and management.
• How It Works:
  • Processes the extracted text to identify key entities and relationships.
  • Constructs a knowledge graph that represents the semantic context of the document data.

3. app.py

• Purpose: Provides an interactive interface for querying the data using a Streamlit app.
• Dependencies: Streamlit for the web application interface.
• How It Works:
  • Allows users to input queries.
  • Retrieves relevant information from the PDF and the knowledge graph.
  • Uses a generative model to provide coherent and contextually relevant responses.

Setup and Installation

1. Clone the Repository:

   git clone https://github.com/HarshMN2345/HybridRAGapproach
   Activate python env

2. Install Dependencies: Ensure you have Python 3.7+ installed. Then install the necessary Python packages:

   pip install -r requirements.txt

3. Run the Scripts:

   • Extract Text:

     python extraction.py

   • Create Knowledge Graph:

     python graphcreation.py

   • Start the Streamlit App:

     streamlit run app.py
     I AM RUNNING OLAMA LOCALLY ON MY PC

Usage

1. Text Extraction: Run extraction.py to process your PDF files and extract text.
2. Knowledge Graph Creation: Execute graphcreation.py to build and populate the knowledge graph.
3. Interactive Querying: Launch app.py using Streamlit to start the web interface. Enter queries in the app to interact with the document data and get responses.

Benefits

• Enhanced Accuracy: Combines retrieval with knowledge graph-based contextual understanding for more accurate responses.
• Contextual Understanding: Utilizes the knowledge graph to provide deeper semantic insights and improved query handling.
• Improved Interaction: Offers a natural and intuitive way to interact with complex document-based information.

Use Cases

• Customer Support: Answer detailed queries about product manuals or legal documents.
• Research Assistance: Aid researchers in extracting and understanding information from scientific papers.
• Educational Tools: Provide students with interactive learning tools based on textbooks and related knowledge.

Feedback and Contributions

For feedback or to contribute to the project, please submit issues or pull requests via the GitHub repository.

License

This project is licensed under the MIT License. See the LICENSE file for more details.

---