Why called Jenny? See here: https://computing.dcu.ie/~humphrys/Turing.Test/08.chapter.html (ref. Jenny18)
Since we were all from outside of Boston, we were curious to see what the residents of Boston felt could be improved on. We reached out to some friends who live in the area and were suggested Boston 311 as a starting point. We discovered that the City of Boston provided many datasets that we could leverage on for our project but we chose to focus on improving the process of filing/reporting non-emergency issues. We noticed that there were four ways to report the issues but the user experience across all those platforms was very different, so we sought to standardize and streamline the process of reporting.
What it does
This app is an attempt to improve the current Boston 311 application. We have an intelligent chatbot named Jenny that listens to the user's requests (i.e. submitting a report) to officially submit the report. Jenny can listen for the location of the report and show the categories of the common issues that people report. The request is submitted and ultimately the collected data in the background would be used to analyze what issues in the neighborhoods of Boston are most critical to look at. The city council would be able to see the data visualization to which issues are most alarming and take action accordingly. In addition, we try to determine the response time to resolve the cases. Have they improved in the last 7 years or so? How long on average does it take to fix an issue under a certain category?
How we built it
We utilized IBM Watson to build the chatbot and integrated with Facebook Messenger. We also have data analysis of the historical 311 data running in another component.
Challenges we ran into
We initially had a lot of ideas that we were bouncing off of, so solidifying our idea to a building a chatbot was one of the first challenges. Moreover, all of us never had any experience using IBM products. We had to learn from numerous hours of struggle on how to set up our accounts and getting our app to run locally. Filtering out the .csv dataset to analyze was also challenging since we did not have much experience with data analysis and using pandas to manipulate the data. We tried integrating the chatbot with Twilio and Facebook, but we could not get Jenny to respond the way we wanted.
Accomplishments that we're proud of
We are still proud that we have at least somewhat of a working product where Jenny actually does talk back to us. In addition, all of us ended up learning something that was not necessarily in the plan, such as Python Flask, Twilio, and Facebook. Twilio and Facebook was implemented slightly last minute, we are more familiar with setting up backend servers and connecting different platforms together.
What we learned
Python, Flask, Pandas, API calls, Leaflet
What's next for Jenny311
The first thing would be to get Jenny to get to work as we had intended to, where she can respond to the user's request. We would then integrate another service to take in the chat data so that the official request could be sent to the database. In addition, we were supposed to utilize the historical 311 data to see if we can predict which type of report (e.g. pothole, graffiti, abandoned vehicle) is most critically reported in specific Boston neighborhoods/districts, but we did not get to that point.