wavelets / laice

Train your own Natural Language Processor from a browser 🤖 (Prototype)

_{Icon by simpleicons}

Train your own Natural Language Processor straight from your browser!

---

What?

Laice allows you to build, train, and classify your own sentences via a Web UI. Laice can also communicate with your applications through a RESTful API.

In other words, laice aims to be a free, open sourced alternative to api.ai, luis.ai, and wit.ai

Laice turns words into meanings (with enough training data), for example:

• Jack lives in Berlin -> {'person': 'Jack', 'location': 'Berlin'}
• Tutu really likes bananas -> {'person': 'Tutu', 'food': 'bananas'}
• Close the door -> {'action': 'close', 'object': 'door'}

---

Getting started

git clone https://github.com/kendricktan/laice.git
cd laice

npm install -g bower
bower install

pip3 install -r requirements.txt

# Optional, doing this will yeild more accurate
# predictions, however will have a slower startup
# time. Its downloading pre-processed data.
python3 -m spacy.en.download --force all 
python3 manage.py migrate
python3 manage.py runserver

Then navigate to http://127.0.0.1:8000 to view your own Natural Language Processor!

---

Contributing

Currently we're using React frontend coupled with a django backend. JavaScript files are written in jsx and then compiled to js using gulp.

Frontend

To get started with frontend development, make sure you have npm, bower, gulp installed globally.

npm install -g bower
npm install -g gulp
npm install
bower install
gulp

Backend

To get started with backend development, make sure you have installed everything in the requirements installed

pip install -r requirements

---