Here in Australia, we are experiencing both misinformation and confusion on what are accurate sources. We felt inspired to build a solution that could be a single source of truth on pertinent information for those that need it,

What it does

  • Updated FAQ
  • Source linking
  • Government health notices
  • Global and national case tracking
  • Helpline assistance
  • Testing centers locations near the user
  • Open/closed status of individual Government schools
  • Latest news links from national broadcasters

How we built it

We used Dialogflow for our natural language understanding solution, which was deployed as a Facebook messenger app. It was also deployed to our website using Komunicate. We sourced our information through web crawlers (BS4 & Scrapy) to build and update our knowledge base. We used cloud functions for both the execution of our web crawlers, as well as our dynamic webhook fulfilment that connects to Firestore and Algolia. We collected and analyzed data collected from the conversation with Google sheets and a data connecter to BigQuery, where we stored this data. We also used Google sheets as part of our continuous deployment pipeline.

Challenges we ran into

  • Effective remote communication
  • Other life commitments/managing burnout
  • Integration between different technologies
  • Setting up an effective continuous delivery pipeline

Accomplishments that we're proud of

  • User acceptance (500 unique users for alpha testing/data collection)
  • Real-time graphical response generation for case tracking
  • Location-based nearest testing center integrated with maps
  • Feedback collection

What we learned

  • User experience is really important in chatbots
  • Rich media is important chatbots
  • Dynamic suggested replies result in more retention
  • Shorter feedback loops speed up develop

What's next for CoronaBot

IVR integration for call centers

Built With

Share this project: