A small test to check how difficult it is to write an intent classifier with an entity extractor.
All of the code was only tested using Python 3.6 so I'd recommend using it as well. Before starting, setup a virtual environment and within that environment, install the dependencies:
pip install requirements.txt
Next, open python up in a shell and download the necessary nltk data as follows:
python
import nltk
nltk.download()
Then you need to download a pretrained GloVe model. I got mine from here. It seems to work really well.
Finally, to run everything just run the following within the virtual environent:
python -m question_answering.main