AI Text Revision System

Access link: http://docassistant.linkpc.net/

A web application for revising and improving text using AI assistance, with multi-tab support and version history tracking.

Features

• Text highlighting and annotation system
• AI-powered revision suggestions
• Multi-tab document management
• Revision history tracking
• Dark/light theme support
• Real-time content updates

Prerequisites

• PHP 8.2 or higher
• Node.js & npm
• Composer
• SQLite (or your preferred database)
• OpenAI API Key

Docekr Deployment Instruction

1. Build and run Docker containers

docker-compose up --build -d

2. Create the database

sudo docker-compose exec db bash
mysql -u root -p
CREATE DATABASE docassistant;

3. Run migrations

sudo docker-compose exec app bash
php artisan migrate

4. Install dependencies

php artisan key:generate
docker-compose exec --user root app bash
rm -rf /var/www/docAssistant/node_modules
npm cache clean --force
chown -R www-data:www-data /var/www/.npm
npm install
npm run build

5. Access the application

Open your web browser and navigate to http://localhost:8090 or the port specified in your docker-compose.yml

Manual Instructions

1. Clone and Install Dependencies

git clone https://github.com/Sergey-Mr/docAssistant.git
cd docAssistant
composer install
npm install

2. Configure the Environment

1. Copy the example environment file and generate the application key:

   cp .env.example .env
   php artisan key:generate

2. Add the following to the .env file:
   • Database configuration: Set up your database connection. For example:

     DB_DATABASE={your_db_name}
     DB_USERNAME={username}
     DB_PASSWORD={password}
     
     DB_CONNECTION=sqlite
     DB_DATABASE=/absolute/path/to/your/database.sqlite

     (Alternatively, configure MySQL or another database of your choice.)
   • OpenAI API Key: Add your OpenAI API key:

     OPENAI_API_KEY=your-api-key-here

3. Set Up the Database

1. Run the database migrations:

   php artisan migrate

4. Start Development Servers

1. Launch the backend:

   php artisan serve

2. Run the frontend:

   npm run dev

Your project should now be available at http://localhost:8000 (by default).

---

Architecture Decisions

Prompt Engineering

The application uses GPT-4 with a specific prompt designed to handle text revisions. The prompt was derived emprically. The model receives the following data:

• Context: the whole text of the user.
• Annotations: the portions of text selected by the user and user`s instructions.

Temperature = 0.7 was derived emperically and proved to be the most suitable level for the purpouse of the web application.

The model is instructed to act as a text revision assistant and performs the following tasks:

• Revise specific annotated text portions while keeping the rest intact.
• Ensure revisions blend seamlessly with the original text.
• Return a JSON response containing:
  1. The full revised text.
  2. Original and revised sections.
  3. Explanations for each change.

Please find more details on prompt implementations: ChatGPT Service Implementation

Backend

• Framework: Laravel.
• Database Structure:
  • tabs: Manages multiple document contexts.
  • text_changes: Records version history with explanations.
  • user_texts: Stores current text content.

Frontend

• Framework: Alpine.js, a lightweight JavaScript library for reactive components.
  • Handles real-time UI updates.
  • Manages tab switching and content state.
  • Minimal overhead compared to larger frameworks.
• Styling: Tailwind CSS, a utility-first framework.
  • Enables rapid UI development.
  • Built-in dark mode support.
  • Ensures consistent component styling.

Key Components

• JavaScript Classes:
  • TextEditor: Core editing functionality with annotation support.
  • TabManager: Handles tab CRUD operations.
  • DashboardNotifications: User feedback system.
• Laravel Controllers:
  • TabController: Manages document organization.
  • TextController: Handles text processing and history.
• External Libraries:
  • Laravel Breeze: Authentication scaffolding.
  • OpenAI API: AI-powered text suggestions.
  • Axios: HTTP client for API requests.
  • AlpineJS: Minimal reactive framework.
  • TailwindCSS: Utility-first styling.

---

License

This project is licensed under the MIT License.

---

Notes

To make the application work:

• Ensure that the OpenAI API key is added to your .env file under the OPENAI_API_KEY variable.
• Make sure your database connection is correctly set in the .env file.