Message author classifier

Dependencies

• NodeJS 6.3.0+
• Python 3.4.3+

Setup

Install SciPy + NumPy by following these instructions.

Install scikit-learn by following these instructions.

Install nltk by following these instructions

0. You now need to download nltk data from nltk package manager.

   Open a python console and run:

   >>> import nltk
   >>> nltk.download()
   

   This will open up a UI from. Locate install 'all-corpora' option and download the corpora.

1. Now add bulgarian stopwords into nltk data.

   Locate nltk download directory (C:\Users\me\AppData\Roaming\nltk_data)

   go into corpora\stopwords

   from project copy people-classifier\processing\stopwords-bg.txt and paste into stopwords folder as bulgarian

Run npm install to download all NodeJS libraries needed.

Start

Make your own secrets.json file using secrets.template.json

Run npm run reload-data to fetch your chat history from Slack API

Run python ./processing/json-to-pkl.py to transform your data into features/labels format, suitable for further processing

Run npm start

You can also run python ./processing/vectorize.py standalone to test the classifier manually from stdin