Skip to content

Chat bot which can be used to answer FAQ questions. Uses cosine vector distance to estimate similarity of the user's question to the FAQ corpus

Notifications You must be signed in to change notification settings

canonicalmg/FAQ-Chat-Bot

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

9 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

picture

Basic FAQ chat bot (Version 1.1)

How to install

Download Python 2.7.x - https://www.python.org/downloads/

Download pip sudo easy_install pip

Download Virtualenvwrapper pip install virtualenvwrapper

Finally,

mkvirtualenv chatbot
pip install -r requirements.txt
python main.py

About the bot

picture Source[https://en.wikipedia.org/wiki/Cosine_similarity]

The bot is trained on the FAQ page for a given site and uses that as it's corpus to compare the similarity between user input and known FAQ questions. If a good match is found it will return the answer, otherwise it will let the user know there was an issue. What determines a 'good' similarity is defined in the Bot.settings object, by default we say that anything over a score of 50% is a good match but feel free to play around with this.

Future work

This will be turned into a back-end service which will communicate with an embeddable front-end widget that can be integrated into any website. It will require that the admin properly configures the corpus so I will probably include a web scraper or give the ability to add your own file.

Back-end will be Django, hosted on Heroku. Using websockets to reply (See https://blog.heroku.com/in_deep_with_django_channels_the_future_of_real_time_apps_in_django) Front-end will be js that initializes communication with the backend and indicates which corpus is to be used. (corpus indication is important because we run the risk of overfitting the data if we do not specify which FAQ page the user is referring to)

About

Chat bot which can be used to answer FAQ questions. Uses cosine vector distance to estimate similarity of the user's question to the FAQ corpus

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages