The problem Consult solves

CONSULT is a responsive android application that maintains a record of blood donors and their blood groups along with requests made by people to the blood donors for emergency blood requirements.

In addition to this, we have added a General Health query related chatbot (Siemens Healthineers Problem Statement). We have first scraped through website to create the dataset and then preprocessed the data in order use it for intents and responses for the Neural Network to learn and reply accordingly.

The chatbot is first trained using Natural Language Toolkit to understand the messages depending upon various tags and randomly choose the appropriate message to reply. We used flask and connected it with twilio in order to receive and send messages through Twilio. The chatbot is then connected to the flask app to use the chatbot on whatsapp to reply to the user. This is done in order to use the same chatbot to reply on whatsapp as well. Hence, our chatbot works for mobile app, website as well as social media messaging apps like whatsapp and telegram on modifying. The flask app is deployed using Heroku on which post requests are made in order ot get a reply back.

Challenges we ran into

The first and main problem we ran into was to find the right dataset for our chatbot. There wasn't any specific dataset which had the required fields. So finally we had the scrape from website. Even for scraping, there were many website with unecessary data and hence it took us time to find the right one and then scrape it out. Apart from the chatbot, some of the issues/challaneges we ran into were the integration of Python Flask API with the android application to make post requests and then display them.