Vitabot

What it does

Our aim is to develop an AI powered chatbot built based on Natural Language Understanding (NLU) which is implemented using the robust Stanford SQuAD Library.
Vitabot scrapes the best rated sites and collects the information related to the user's question and provides it to the NLU model. The model tries to understand the context of the scraped output and provides the desired results to the user.
It enables people to take their screening from their comfort zone, ensuring the safety of the user in these pandemic situtations.
Since the information collected is from top rated medical sites, it provides a trust factor to the user through its answers to their health queries.
We integrated it with a telegram bot for time being, for prototyping purpose.

Challenges we ran into

We faced few bugs while implementing the transformers for the collected search query with Stanford SQuAD model.
It took nearly 8 hours to properly complete the Colab code for the task in hand. So, we found time as a challenging factor during the hackathon.
Also, we struggled hard to fit out RASA model to the Docker container. It was raising a Timeout issue as our model needed GPU from the hosting end.
Since we haven't used Stanford NLU models so far, we had to learn those concepts very quick within the provided time and implement it in our solution.

What we learned

We learnt a lot about the Stanford SQuAD dataset and transformers while implementing NLU (Natural Language Understanding). We learnt about RASA - an open source machine learning framework used for building automated text assistants. We learnt about Docker containers and Telegram bot integration too, during the process.