Hackathons were once an opportunity for developers to build, network, and afterwards enjoy free cocktails. The pandemic robbed us of our free cocktails, but the virtualized 2020 Microsoft Encode hackathon let us enjoy the building and networking all the same.
The Hackathon Theme
Four major categories of high-level problems were presented:
· Financial Services
After some deliberation, we chose the Government theme with its scenario that outlined the following challenge: “create something using crowdsourced data to identify, categorize, and track problematic area city services… and address accessibility”, moreover, “build a feedback loop for citizen engagement and trust between government and citizen.”.
After gathering domain knowledge from a 311 operator in Toronto, we realized that a feedback loop (between citizen and local government) does not exist for those raising real issues and complaining online.
Scrolling through tweets to @cityOfToronto highlights an accessibility issue. Citizens are often redirected to the Toronto 311 phone line after tweeting a complaint, but why do they need to call 311 if they’ve already submitted a complaint online?
We built a system using Azure Cognitive Services (on top of a python flask app) that analyses sentiment, recognizes key phrases, and categorizes input text (such as a tweet). The result is a system that assists the difficult job of a 311 operator, without replacing them.