I built a ChatBot that communicates via Twilio's API on WhatsApp. The Bot gives Covid-19 statistics from the US & the UK. It also provides some positive news articles to read in light of recent times.
This is a simple ChatBot that communicates via Twilio's API on WhatsApp. The Bot gives Covid statistics from the US & the UK. It also provides some positive news articles to read in light of recent times.
How it works
Users interact with WhatsApp to communicate with the ChatBot
The Bot replies to greetings along with different variations of the following questions.
Sample Questions:
Input
Output
Greeting!
Hi
Hi, How is it going?
Good
uk confirmed cases?
Confirmed Uk Cases 165221
How many uk confirmed cases?
Confirmed Uk Cases 165221
What are the us covid stats?
United States Stats: confirmed: 1055303 recovered: 144423 critical: 18665 deaths: 61112
Automated responses and bot training with Chatterbot
Some Issues
Setting up ChatterBot produced issues with the nltk package.
It took some time to find the right way to set up deployment in Heroku, FastApi is a relatively new Python framework so there are not as many resources for it as Django or Flask.
Querying news articles based on keywords did not always produce 'good' news, however, I selected The Guardian API because you can also query on basic logic operators.
The Chatbot doesn't always return the desired results.
Things I learned
Since this is my first Hackathon and I am quarantined in another country I decided to use the time and opportunity to learn all new things! This is the first API I've created with python. I decided on FastApi because of its simplicity and great features (especially the built-in docs 🙌🏾).
I also wanted to add a machine learning component so after some research, I landed on ChatterBot. I think it is a great introduction for someone wanting to learn some of the basics of ML.
Additional Resources/Info
I used the Guardian API to get news articles and a Covid-19 API from rapidapi.com to get coronavirus statistics.
Ari 🪐
