License

⚠️ This repository is for classroom demo only. Please do not reuse or fork this repo.

Personal Knowledge Base

A local-first personal knowledge base application that integrates document management, vector embedding, and AI-powered Q&A capabilities.

Features

• Document Management: Upload, categorize, and organize documents in a multi-level folder structure
• Google Drive Sync: Automatically sync documents with Google Drive for backup and accessibility
• Vector Search: Convert documents into semantic vectors for efficient retrieval
• AI Question Answering: Ask questions about your documents in natural language
• Privacy-First: Run locally with your own models, keeping your data private

Getting Started

Prerequisites

• Python 3.10+
• Docker (optional, for running Ollama)
• Google account (for Drive integration)

Installation

1. Clone this repository

git clone https://github.com/yourusername/personal-knowledge-base.git
cd personal-knowledge-base

2. Create a virtual environment

python -m venv venv_kbs
source venv_kbs/bin/activate  # On Windows: venv_kbs\Scripts\activate

3. Install dependencies

pip install -r requirements.txt

4. Set up Google Drive API

   • Create a project in Google Cloud Console
   • Enable Google Drive API
   • Create OAuth 2.0 credentials
   • Download the credentials as client_secret.json and place in the project root

5. Configure environment variables by copying the example

cp .env.example .env

6. Set up Ollama (for local LLM)
   • Install Ollama or use the Docker configuration
   • Pull required models:

ollama pull mistral:7b-instruct
ollama pull bge-m3:latest

Running the Application

Option 1: Local Development

streamlit run main.py

Option 2: Using Docker

docker-compose up -d

Configuration

Edit the .env file to customize your setup:

• OLLAMA_URL: URL for your Ollama instance (default: http://localhost:11434)
• DRIVE_FOLDER_ID: Google Drive folder ID for document storage
• VECTOR_DRIVE_FOLDER_ID: Google Drive folder ID for vector database backup
• USE_WEB_SEARCH: Enable/disable web search for question answering

Architecture

• Frontend: Streamlit-based UI for interaction
• Document Processing: Converts documents to vectors using embedding models
• Vector Database: ChromaDB for efficient semantic search
• LLM Integration: Local Ollama models (default: Mistral 7B)
• Storage: Local files with Google Drive synchronization

Deploying to Streamlit Cloud

1. Push your code to GitHub
2. Connect your repository in Streamlit Cloud
3. Configure the following secrets in Streamlit Cloud:
   • google_creds: Content of your Google credentials JSON file
   • Add any additional environment variables from your .env file

Contributing

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

License

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

Project Structure

knowledge-base-assistant/
├── components/        # UI components
├── core/             # Core business logic
├── services/         # External services integration
├── utils/            # Utility functions
└── data_base/        # Knowledge base storage

Requirements

• Python 3.10+
• Streamlit
• LangChain
• (Optional) Ollama for local AI models

Notes

• Files are stored in categorized folders
• Supports multiple document formats
• Web search is optional and can be enabled/disabled
• All chat history is session-based and not persisted