Using data science to improve society is one of the biggest ethical trends in computing. However, it’s often used to increase efficiency — without asking what an individual community really needs. Georgia Tech researchers want to not only help communities, but involve them.
“When you’re doing careful data science, you do care about efficiency, but there are a lot of other values like residents’ voices in the data science process and then in the outcome of what the data gets used for,” said School of Computer Science (SCS) research scientist Amanda Meng.
Meeting a community
For three years, Meng and School of Literature, Media, and Communication Associate Professor Carl DiSalvo have worked with the Westside Atlanta Land Trust, an initiative of the nonprofit, HELP ORG, Inc., that has been trying to establish a community land trust. A community land trust is a nonprofit that owns the land while residents buy the buildings, enabling more accessibility and affordability. Yet before they can become a fully functioning community land trust, they must survey the land to catalog conditions of structures and vacant lots. In March and April 2017, a team of residents collected this data and plotted it on a map, with aid from Meng and SCS master’s student Tom Nguyen.
Having residents collect data is not only empowering for their community, but also leads to more accurate collection. It’s one of the fundamental concepts of Meng’s research with SCS Professor Ellen Zegura and DiSalvo on bringing the sociological concept of care — an attentiveness to the desires and values of the community — to data science.
“The residents know the area,” Meng said. “They are the best at collecting data on what’s changed.”
Working with a community
Collaborating with a community can be challenging for both parties. Researchers must build a relationship so the community trusts them. Meng has been attending community meetings for years and communicates extensively to make sure everyone understands the process.
Yet researchers also must rely on residents to attend data collection meetings, while meeting grant and paper deadlines. Researchers and residents also often have to collaboratively tinker with the data to ensure it’s as clean as possible, from matching street names to what’s in the Fulton County database to checking that each data point corresponds accurately to the map. The effort is worth it, though.
“It’s so rewarding being invited to walk the street with community members and being entrusted with this important goal of capturing changes in property conditions and values,” Meng said.
Meng, along with Zegura and DiSalvo, presented their paper, Care and Practice of Data Science for Social Good, at the new ACM Conference on Computing and Sustainable Societies (COMPASS 2018) held in California from June 20 to 22. The new sustainable society and computing conference was co-organized by Zegura to bring together similar disciplines like computational sustainability and data science for social good.