The system takes a description of the user’s issue and suggests the case types that are most likely to fit the description.
Boston has launched a new crowdsourcing effort to gather data for its 311 interface, deploying a machine-learning model that takes a description of the user’s issue and then suggests case types that are most likely to fit what they need.
It sounds like a complex concept, but it stands to make the city’s 311 program much simpler. The way 311 works now is that it gives users a list of dozens of the most frequent requests. The user must then sort through the items on this list to find his or her concern before being directed to a relevant form. The new system will simply ask users to describe what’s wrong in a sentence or less. The computer then offers a much shorter list of potential case types, resulting in a quicker and easier process for residents.
“The end goal of all this is really just to provide an experience for the users where it’s much, much easier for them to get where they’re going,” said Andrew Therriault, Boston’s chief data officer, “for them to provide the right information to use in an easy fashion so that people find the system much easier to use and are more likely to report things that we can do to help.”
The way the model is built, it will become more effective once the public uses it more, because it will be able to better predict what exactly they’re asking for based on the language they use. The system will collect data that links certain keywords and phrases to certain case types. For example, if people keep saying, “no one has cleaned the snow off my sidewalk,” the program will connect it to a case type for “sidewalk not shoveled,” putting that option at the top of the list.
This could prove especially useful once the system starts to identify patterns. Going back to the snow example, the model might start to notice that a high volume of snow requests is coming in during a blizzard, and so it would move to automatically put snow options at the top of its list until the deluge of related requests fades.
Therriault said that Boston technologists expect this to be a starting point for similar platforms, ones that can provide the same service requests sent via email and text, potentially even streamlining requests for things like trash pickup or pothole repair directly into a service queue. Such tools would make government more efficient and give citizens the option to engage easily through the channels they are most comfortable using.
In a press release announcing the platform, Boston Mayor Marty Walsh emphasized that this is the latest in the city’s efforts to use data to make life for its citizens easier. Boston has been working hard this year to use tech, transparency dashboards and tools driven by open data to engage with residents, helping them do things such as easily find rental history for properties or check the status of governmental progress toward city-planning goals.
This innovative 311 tool is another part of all that, and the city plans to release its data and model code on its Analyze Boston open data hub, so that other jurisdictions can potentially use it to create their own systems.
Constituents can file reports through the 311 website, 311 mobile application, or by tweeting to @BOS311. Citizens who wish to help the city gather more data to improve the new 311 system can help out by visiting this Web app.