Inspired by the rise and relevance of voice assistants like Google Home and Amazon Alexa, we were motivated to develop a consolidated and accessible platform through which anyone could receive relevant and reputable information on the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), or COVID-19. The exponential growth of COVID-19 not only marks the importance of how soon we all act, but also the importance of spreading correct and fact-checked information. The target demographic for this voice application includes elderly people who may not be as accustomed to using modern technology (or would like to refrain from physical contact with devices such as phones), as well as people with disabilities. Our application, COVID-19 Hub has not currently been launched for everyone to use and is still in beta stages, but it very elegantly shows the beauty of modern NLP technologies such as Google's Dialogflow, making it a little bit easier for everyone to stay up-to-date with the whirlwind of information that surrounds us.

What it does

COVID-19 Hub was designed with the intent to inform and educate, providing current statistics (including week-by-week changes to indicate the flattening of the curve), tweets from reputable organizations like the WHO or CDC, as well as actionable advice for those who think they may be developing symptoms. Invoked through voice commands, our application is responsive and conversational, assisting the user in navigating closer towards factual information without having to worry about the reputability of most sources on the Internet.

How we built it

Built using Google Dialogflow with custom fulfillment, the majority of our app's processing takes place using the Javascript invoked on the backend, using Cloud Functions for Firebase to handle external API requests. Certain portions of our app, such as providing synonymous invocations of intent ("get tweets" vs. "what are people saying on twitter") were created using Python and its well-documented NLTK library. This made the entire process of providing synonyms to Dialogflow a lot easier. This was done by scripting separately and using a Flask webhook to match intents to their synonymous phrases.

Challenges we ran into

Neither of us had previously used the Google Cloud platform for a medium-scale project such as this one, so we had to go through tons of Dialogflow documentation to understand how to create custom intents, how to use Firebase to deploy our webhook code, and how to perform external API calls in our fulfillment requests. Krish was also new to hackathons and coding, so it was a challenging yet rewarding process getting him up to speed with the technologies used in this project.

Accomplishments that we're proud of

For starters, we both love to use our voice application, as it saves us a ton of time and effort Googling the information, scrolling through Twitter, or looking for relevant news articles. After introducing the voice app to our parents, they also really enjoyed its simplicity and ease-of-use, asking us both to add this app to all the Google Home Minis in our house. Now, no matter where we are, we can always receive updated and fact-checked information on COVID-19.

What we learned

How to design a conversational voice agent, how to mimic realistic conversation flows, how to use Google Dialogflow.

What's next for COVID-19 Hub

Implementation with Google's NoSQL Firebase datastore to cache information on symptoms and disease prevention, saving us valuable time in performing API calls on information that we already have access to. We were also thinking of providing updates on vaccine development and financial movements in the stock market in response to the increased spread of the virus.

Share this project: