As illustrated in Robert Putnam’s renowned book “Bowling Alone: The Collapse and Revival of American Community,” Americans have become increasingly isolated over the decades, often spending leisure time alone without social gatherings. During the COVID-19 pandemic, this issue of isolation was exacerbated, calling further attention to the public health crisis of loneliness and isolation in the United States.
To help encourage in-person gatherings, Event-Based Social Networks (EBSNs), such as Meetup.com and Facebook Events, have become an increasingly vital tool for facilitating these occasions based on shared interests — ranging from farmers’ markets to game nights. To maximize the effectiveness of EBSNs, a group of Mason faculty members with interests in community engagement, machine learning, and geographical data analysis wanted to take a closer look at how these arranged local gatherings fluctuated depending on community and group characteristics. They were able to undertake this analysis following the approval of their 4-VA@Mason Collaborative Research Grant proposal entitled “AI for AI: Toward Community-level Human-AI Collaborations in Local Meetups.”
Led by Myeong Lee, Mason’s Assistant Professor of Information Science and the Director of the Community Informatics Lab, the researchers also included former College of Science faculty members Olga Gkountouna, who assisted with machine learning model development, and Ron Mahabir who provided insight on geographical data analysis. Amr Hilal of Virginia Tech helped with data analytics from a machine learning perspective.
While it is known that EBSN users’ participation in Meetup events are influenced by group organizers’ promotions and event frequency, the effects of ecological factors, such as the number of similar groups surrounding a Meetup group, had not been previously studied. The goals of the project were to quantitatively examine how EBSN groups’ ecological features shape the performances of Meetup groups within that organizational ecology. They also wanted to create baseline benchmarks for how state-of-the art AI technologies can predict Meetup groups’ success.
To do so, the team conducted two studies of Meetup data for 500 cities in the US, extracting factors pertaining to “Meetup niches,” which considers similar groups surrounding a Meetup location.
The results revealed intriguing patterns, one of which was that if a Meetup group’s description resembles other groups 
in their geographical area, it tends to attract more participants. In a second finding, the team implemented three advanced machine learning models to predict the success of local Meetup groups, finding that the performances of these prediction models vary across different categories and cities, with some outperforming the state-of-the-art models.
“Overall, our research during the 4-VA project period will provide a basis for understanding human-AI collaboration at the community level by revealing how various factors shape and predict the success of local groups,” says Lee.
Lee credits the success of their findings to a strong team of student researchers, including graduate students Julia Hsin-Ping Hsu who worked on developing deep learning models and ecological features and Ishana Shinde who assisted in calculating community-level features. Undergraduates Victoria Gonzales focused on descriptive statistics of variables; Joel Adeniji managed visualization; and Nnamdi Ojibe handled data cleaning and geographical data aggregation.
The group is now disseminating their findings in the field – one study was published at the International Conference on Communities and Technologies (C&T), and the other is under submission to a premier journal. Lee is planning to write an external NSF grant using the preliminary results from the research, proposing the curation of Meetup-based social gathering data with the promising community-level ecological factors.
“The 4-VA@Mason grant significantly helped me and my team jump-start the project and develop the research studies,” says Lee. “What’s more, it allowed the team to connect with researchers outside of Mason to discuss additional meaningful community-based topics, thus broadening our future possibilities.”