As communities grow and technology advances, planners of “smart cities” have turned to social media as an additional way of garnering residents’ feedback.
However, the sheer volume of content from those platforms can be a challenge for municipal officials looking for meaningful dialogue.
Researchers at the ÁůľĹÉ«ĚĂ’s are hoping to remove that barrier with the help of artificial intelligence and machine learning.
Collecting data from X (formerly Twitter) and building analysis models and visual datasets with the help of AI, Schulich master’s student Mitra Mirshafiee is looking to help decision-makers understand the emotions and context of each post.
Along with , PhD, and , PhD, Mirshafiee has been working through posts about The City of Calgary’s infrastructure and planning to identify issues and differences in opinion on what’s important.
“What our work represents is that it’s not enough to just monitor how the physical facilities are doing,” she says. “The city should also take into account what people are concerned about, especially when it comes to long-term planning.”
A variety of opinions
Smart cities are described as municipalities that aim to improve quality of life for everyone by integrating technology and data solutions to improve operational efficiency and information sharing.
While Calgary is a “smart city,” its combination of the 311 phone and online system, along with focus groups, to determine what is more important to residents doesn’t capture everyone and is viewed as laborious.
“The issue with the calls is that they require too much human intervention and revolve around physical system deficiencies in the city. Hence, they don’t have enough information about the opinions of people on City plans and strategies.”
Mirshafiee points to younger generations, who don’t pay attention to surveys, don’t attend workshops or community meetings, or even bother to call The City to complain about programs.
Another gadget in the toolbox
In a she wrote with Barcomb and Tan, Mirshafiee indicates they looked at a variety of topics on what was at the time called Twitter related to Calgary and started narrowing down the amount of content to sort through.
She says the AI-directed filter was used to capture keywords and query definitions to help include or exclude certain pieces of information.
Mirshafiee says they then used hierarchical algorithms to identify unimportant or unnecessary posts like spam or bots, before identifying and classifying each of the topics and overarching emotions being felt.
For the analysis, she randomly chose 160 posts and manually labelled them as having an emotion of anger, joy, optimism or sadness.
The topics ranged from downtown redevelopment plans and housing prices to seniors’ care and even the Stampede.
“For example, when exploring the topics, we see many people complaining about the money spent on developing one specific part of the city or on special events for specific parts of the community,” Mirshafiee says. “These are not things that they can or are willing to talk to 311 agents about.”
Mirshafiee says social media commentary is another tool for communities to gather feedback and shouldn’t be used to replace any of the existing strategies.
More-informed decisions
With her , Mirshafiee is hoping to connect with municipal and provincial officials to see how they might benefit from the data and the approach to collecting it.
“We saw how residents are voicing their concerns, complaints and suggestions through these comments, and knowing about these concerns may be very helpful to the local authorities as they decide what to do next,” she says.
Mirshafiee says new AI technologies — as well as surveying additional social media platforms — might also make the picture clearer for decision-makers.
For now, she is focused on building an application that will showcase the usefulness of what she has done to this point.
