gilbeckers / tensorchat_fb

A simple context-aware facebook messenger chatbot built using tensorflow. Training of the model is done offline and deployed as a Flask app on Heroku.

Facebook Messenger Bot

This module is a chatbot developed using python and tensorflow backend. The original concept is adapted from here: https://blog.hartleybrody.com/fb-messenger-bot/. The module can be integrated with facebook messenging api by deploying onto Heroku app server. Training of the intents is done offline and later deployed into Heroku cloud.

Dependencies

• Flask v0.11.1
• Jinja2 v2.8
• MarkupSafe v0.23
• Werkzeug v0.11.10
• click v6.6
• gunicorn v19.6.0
• itsdangerous v0.24
• requests v2.10.0
• nltk v3.2.4
• tensorflow v1.2.0
• numpy v1.13.1
• tflearn v0.3.2
• h5py v2.7.1
• python v3.5.2

Usage Detail

Modify the intents.json file to add or remove your response for the bot, To train:

   $ python3 train.py

For evaluation:

   $ python3 eval.py

For Deployment onto facebook messenger

follow the instructions at: https://blog.hartleybrody.com/fb-messenger-bot/