The four creators of Livehoods.org have made a website that shows the character of different neighborhoods in large cities by using check in data from the social media app foursquare to train machine learning.
The website highlights popular venues and ties them together, providing an interesting look on where people go to eat, learn, listen to music, or any number of other common activities.
Livehoods.org was created by Carnegie Mellon University Ph.D. students Justin Cranshaw and Raz Schwartz and professors Jason Hong and Norman Sadeh.
“We’re very interested in the idea of social media, big data and cities and what we can learn about cities,” said Schwartz, who is a visiting scholar from Bar-Ilan University in Israel. “We started (thinking) about this project and about what sort of insights we can learn from social media, and the result was Livehoods.”
Cranshaw, Schwartz, Hong and Sadeh took the more than 18 million data points from foursquare’s API, trained their machine learning algorithms with the data and overlaid it on top of a Google map. The result is a colorful, interactive site packed with information. Currently the website has maps for Pittsburg, where Carnegie Mellon is located, New York City and San Francisco.
“There is so much local knowledge that’s really valuable that’s stored in cities,” said Cranshaw. “What we wanted to do is see if we could extract that local knowledge through social media. People are very excited about it. I think it’s an idea that’s very intuitive, and people have seemed to really latch onto that.
Cranshaw said the project has drawn interest from real estate developers, city planners and professionals “looking for these methods to give you automatic insights of what’s going on” at the ground level.
He predicted the information could also be useful for business owners looking to diversify their offerings, franchises evaluating where to place a new locations.
“There are so many offerings going forward,” he Cranshaw said.
Schwartz said the group didn’t have many decisions to make in terms of presentation; the color scheme is what’s available from Google. He also said there are inherent biases in the data because it’s from a single source, but that foursquare had a broad range of users. According to the Pew Research Center, nearly one in five adult smartphone owners “checks-in” with a location-based service, and one in 10 adults overall checks-in.
But the result has been popular, with users spending significant time exploring the three available city maps. Livehoods.org is asking for users to vote on Facebook what city to map next; as of May 14, Chicago led the pack of more than 90 choices.
“What we wanted to achieve was a way for people to connect with the data and to understand it in an intuitive way,” Schwartz said. “The responses that we got are that people can really relate to it, and that people spend hours clicking on the various neighborhoods.”
Schwartz and Cranshaw are scheduled to present their academic paper at the Sixth International AAAI Conference on Weblogs and Social Media at Trinity College in Dublin.
Email Staff Writer Ian B. Murphy at email@example.com.