Covid-Stats-ChatBot

ChatBot that gives real time statistics for any location in India. It currently accepts any post-office/district/state name.

Bot utilises the following API's:

Covid19india API:

• https://api.covid19india.org/data.json
• https://api.covid19india.org/state_district_wise.json

Postal PIN Code API:

• http://www.postalpincode.in/Api-Details

About

Project Overview

A bot that gives real time covid-19 data can be very handy. It also handles general chitchat and maintains an interactive conversation.

The main tasks involved in this project are as follows:

• Defining various intents that we want the bot to identify
• Defining the domain, general stories and rules that it needs to follow
• Defining actions that it needs to perform in various circumstances
• Training the model on this data

Demo

How to use

• Clone the repo and change your directory:

  git clone https://github.com/motua16/Covid-Stats-ChatBot

  cd Covid-Stats-ChatBot

• Create a virtual Environment and activate it

  python3 -m venv env

  source env/bin/activate

• Install rasa-x

  pip install pip rasa-x -i https://pypi.rasa.com/simple

• Open rasa shell and interact

  rasa shell -m models/final.tar.gz

Features and Limitations

The bot currently has these features:

• It remembers your name from earlier in the conversation
• It gives the data for district corresponding to the pincode/city/post-office name you enter. For eg: if you type 110016, which is the pincode for Hauz Khas, New Delhi then it gives you the data for the District under which it comes i.e., South West Delhi
• It asks if you want previously calculated results, if prompt is yes: then the previous result is shown, otherwise Nationwide data is shown
• It can handle basic greetings and conversations

These are the limitations:

• Since there can be multile places in India with the same name, sometimes we get results that are not what we expect. So it is better to go with pincodes
• There can be minor spelling difference for the same place in the 2 API's that are being called, in which case we dont get desired results. for eg Visakhapatnam and Vishakhapatnam
• Sometimes the bot can interpret name of a person as a place or vice versa. So it is better to use complete sentences like My name is John when asked

Repository Structure

- bot
| - actions # package containing actions
| |- __pycache__  #bytecode-compiles version of program files
| |- __init__ # defines a package
| |- actions_1.py # handles calls to api's
| |- actions_2.py # handles name of user
| |- actions_3.py # handles if the user wants previously fetched results

| - data
| |- nlu.yml # intents definition
|-|- rules.yml  # defines rules
|-|- stories.yml # defines stories

- demo
|- gif.gif # animation 

- models
|- final.tar.gz # final trained model

- README.md

- tests
|- test_stories #  test stories

- config.yml # configuration
- credentials.yml # adding REST endpoints
- domain.yml # define domain of the bot
- endpoints.yml # defining action endpoint

Further Steps

• Deploy this bot on WhatsApp or Telegram
• Deploy as an online application
• Giving more training examples and defining strategies so that the bot can effectively distinguish between name of a person and name of a place.