Inzeera/Multi_Document_Summarizer_Plus_QA_Bot

Features

• —Upload PDF, DOCX, XLS, and XLSX files
• —Automatic document parsing
• —Intelligent text chunking
• —Semantic search using embeddings
• —Retrieval-Augmented Generation (RAG)
• —AI-powered document summarization
• —Question Answering with source citations
• —Download summaries as TXT or PDF
• —Modern glassmorphism-based user interface
• —Supports multiple document uploads

Architecture

User Uploads Documents
          │
          ▼
Document Loaders
(PDF / DOCX / XLSX)
          │
          ▼
Text Chunking
          │
          ▼
HuggingFace Embeddings
          │
          ▼
Chroma Vector Database
          │
          ▼
RetrievalQA Chain
          │
          ▼
Groq LLM
          │
          ▼
Summary / Answer

Project Structure

document_qa_bot/
│
├── app.py
│
├── config/
│   └── settings.py
│
├── loaders/
│   ├── pdf_loader.py
│   ├── docx_loader.py
│   ├── excel_loader.py
│   └── document_loader.py
│
├── rag/
│   ├── embeddings.py
│   ├── llm.py
│   ├── vectordb.py
│   ├── qa_chain.py
│   └── summarizer.py
│
├── utils/
│   ├── hash_utils.py
│   └── pdf_export.py
│
├── temp_docs/
│
├── requirements.txt
│
└── README.md

Technologies Used

Frontend

• —Streamlit

AI Framework

• —LangChain

Vector Database

• —ChromaDB

Embedding Model

• —sentence-transformers/all-MiniLM-L6-v2

Large Language Model

• —Groq API
• —openai/gpt-oss-120b

File Processing

• —PyPDF2
• —python-docx
• —pandas

PDF Generation

• —ReportLab

Installation

Clone Repository

git clone https://github.com/your-username/document-qa-bot.git
cd document-qa-bot

Create Virtual Environment

python -m venv venv

Activate:

Windows

venv\Scripts\activate

Linux / Mac

source venv/bin/activate

Install Dependencies

pip install -r requirements.txt

Environment Variables

Create a .env file in the project root.

GROQ_API_KEY=your_groq_api_key_here

Running the Application

streamlit run app.py

The application will open in your browser.

Workflow

Document Upload

Users upload one or more files:

• —PDF
• —DOCX
• —XLS
• —XLSX

Document Processing

The system:

1. 1.Extracts text
2. 2.Creates LangChain documents
3. 3.Splits text into chunks
4. 4.Generates embeddings
5. 5.Stores vectors in ChromaDB

Summarization

The uploaded document content is summarized using a Groq-hosted Large Language Model.

Question Answering

User questions are:

1. 1.Converted into embeddings
2. 2.Matched against document embeddings
3. 3.Relevant chunks are retrieved
4. 4.Groq generates answers based on retrieved content
5. 5.Source citations are displayed

Example Questions

• —What is the main objective of the document?
• —What conclusions were reached?
• —Summarize the methodology section.
• —What are the key findings?
• —List the recommendations mentioned.

RAG Pipeline

Document
   │
   ▼
Chunking
   │
   ▼
Embeddings
   │
   ▼
Vector Database
   │
   ▼
User Query
   │
   ▼
Query Embedding
   │
   ▼
Similarity Search
   │
   ▼
Relevant Chunks
   │
   ▼
LLM
   │
   ▼
Answer + Citations

Performance Optimizations

• —Cached embedding model
• —Cached LLM initialization
• —Session state management
• —Chunk-based retrieval
• —Efficient vector search

Future Enhancements

• —OCR support for scanned PDFs
• —Conversation memory
• —Multi-document comparison
• —Hybrid search (BM25 + Vector Search)
• —Citation highlighting
• —User authentication
• —Cloud deployment
• —Database-backed document storage

License

This project is intended for educational and portfolio purposes.

Author

Developed by Inzeera Z