Webtools

The ingredient parser library provides a convenient web interface that you can run locally to access most of the library’s functionality for testing and development purposes. The web app (webtools) supports basic workflow needs, namely:

• Testing and using the trained parser (a.k.a. parse_ingredient)

• Browsing the ingredient sentence training database (see Explanation/Training data)

• Editing the labels of ingredient sentence training entries (see Explanation/Training data)

• Training and tuning the parser (a.k.a. train_single, train_multiple, grid_search)

  • train_single trains a model using the provided hyper-parameters and outputs the model accuracy statistics.

  • train_multiple trains multiple models using different random seeds values and outputs the aggregated accuracy statistics.

  • grid_search trains multiple models using different hyper-parameters and outputs the accuracy statistics for each set of hyper-parameters.

Requirements and Setup

To run the webtools, clone the repository and install all python libraries in requirements-dev.txt. Next, install Node, a Javascript runtime, on your machine. Lastly, navigate to the webtools directory and install using npm:

# Clone repository
$ git clone https://github.com/strangetom/ingredient-parser.git
$ cd ingredient-parser

# Create venv and install required packages
$ python -m venv venv
$ source venv/bin/activate
$ python -m pip install -r requirements-dev.txt

# Install webtools
$ cd webtools
$ npm install
$ npm run dev

The webtools can be accessed at http://localhost:5000 in your browser.

Tech Stack

The web technology stack (packages, libraries, and tooling) includes:

• Typescript

• React

• Mantine

• Zustand

• Vite, BiomeJS

• Socket.IO

• Flask

• Flask-SocketIO

• Flask-CORS

The list is subject to change as the webtools evolve to support changing needs or new functionality.