DocMind AI

DocMind AI is an intelligent document question-answering system built using a Retrieval-Augmented Generation (RAG) pipeline. Users can upload PDF, DOCX, PPTX, or TXT files and ask natural-language questions about the document content.

Features

• Upload and analyze PDF, DOCX, PPTX, and TXT files
• Text extraction and document processing
• Semantic search using FAISS vector database
• Sentence Transformers embeddings
• Local LLM answer generation using Ollama and Llama 3
• Context-aware question answering

Technologies Used

• Python
• Streamlit
• FAISS
• Sentence Transformers
• Ollama
• Llama 3

WorkFlow

flowchart TD
    A[Document Upload] --> B[Text Extraction]
    B --> C[Text Chunking]
    C --> D[Embedding Generation<br/>Sentence Transformers]
    D --> E[FAISS Vector Store]
    E --> F[Similarity Search]
    F --> G[Llama 3 via Ollama]
    G --> H[Answer Generation]

Loading

Project Structure

docmind-ai/
├── app.py
├── src/
├── assets/screenshots/
├── requirements.txt
└── README.md

Installation

pip install -r requirements.txt

Run the Application

python -m streamlit run app.py

Screenshots

Home Page

Document Upload

Question Answering

Limitations

• The application works best with text-based PDF, DOCX, PPTX, and TXT documents and may not perform optimally with image-based or non-text content.
• Very large documents may require more memory and longer indexing time.

License

MIT License