An AI powered web-app for estimating language proficiency expectations of adult English learners.

The application can be accessed at the following URL:

https://language-proficiency-estimator.netlify.app

To run the application locally for evaluation or development purposes:

1. Start the backend server
2. Start the frontend server
3. Visit http://localhost:5173 in a web browser

Requires Python 3

The backend server requires the following python packages to be installed: (See backend/requirements.txt)

• Flask
• torch
• flask_cors
• scikit-learn
• matplotlib
• numbpy

These may be installed by using pip / pip3

pip3 install Flask torch flask_cors scikit-learn matplotlib

Or in Unix-like environments, they may also be installed using venv from the backend directory.

python3 -m venv .venv
source .venv/bin/activate
python3 -m pip install -r requirements.txt

Once the dependancies have been installed, the server may be started by running the following command from the backend directory.

flask run

Node (v. 18 or greater) and npm are required. From the frontend directory run the following:

npm install

npm run dev

The full working application should now be accessible at http://localhost:5173

The backend can be deployed using Fly.io. To deploy, install flyctl and then from the backend directory run:

fly deploy

Building the frontend for production deployment requires the following environment variable to be set:

• VITE_PROFICIENCY_API - URL for the backend proficiency model API (defaults to http://localhost:500)

To build the frontend for deployment run the following command from the frontend directory:

npm run build

The frontend/dist directory contains the packaged front-end files for deployment.

This repository includes a pre-trained model at backend/model.pt. But you can also train it by running the following from the backend directory.

python3 train.py

The model has been trained on the BEST Dataset of Language Proficiency. Below are some visualizations of the data from this dataset.