Disclaimer: This project was created in around 6 hours for a hackathon with Orano. As I was discovering many technologies during this process, the code should be considered as a proof of concept rather than production-ready code.
T-REX is a Progressive Web Application (PWA) designed to manage and search through "Fiches de retour d'experience" (Experience Feedback Forms) for Orano. The application combines modern web technologies with machine learning to provide an intelligent document management system.
- Create and submit experience feedback forms
- Rich text editing capabilities using CKEditor
- Automatic text embedding using BERT model
- Offline support through Service Workers and IndexedDB
- Vector-based semantic search using BERT embeddings
- Multiple filtering options:
- Keyword search
- Name filter
- Location filter
- Date filter
- Intelligent content and title matching
- Download submitted forms as PDF documents
- Formatted for easy reading and sharing
-
Frontend:
- HTML5, CSS3, JavaScript
- Progressive Web App (PWA) capabilities
- CKEditor for rich text editing
- Service Workers for offline functionality
-
Backend:
- Node.js with Express
- TensorFlow.js for BERT embeddings
- SQLite3 for data storage
- PDFKit for PDF generation
-
Key Dependencies:
- @tensorflow-models/universal-sentence-encoder
- @tensorflow/tfjs
- express
- pdfkit
- sqlite3
The main interface for creating new experience feedback forms with rich text editing capabilities
Confirmation message shown after successful form submission
Interface for searching and viewing existing experience feedback forms with multiple filter options
Example of an exported experience feedback form in PDF format
- Clone the repository
- Install dependencies:
npm install- Start the server:
node src/server.jsThe application includes offline support through:
- Service Workers for caching static assets
- IndexedDB for storing documents when offline
- Background sync to upload cached documents when connection is restored
Given the hackathon time constraints, several areas could be improved:
- Enhanced error handling
- Better test coverage
- Improved UI/UX
- Security enhancements
- Code optimization
- Add proper npm scripts in package.json