Goggins is an assistant professor in the iSchool with an interest on the uptake and use of information and communication technologies by small groups.
When news breaks on Twitter, it’s easy to let 140 characters after 140 characters disappear into the depths of the constantly refreshing news feed.
It might be an inconvenience for tweeters hoping to catch every last bit of commentary, but it poses an even bigger problem for scientists looking to study patterns in social media.
Enter TwitterGoggles, a program—developed by doctoral student Alan Black, master’s student Michael Gallagher and iSchool College of Information Science and Technology associate professor Sean Goggins—that collects and saves tweets on a given topic to be analyzed by a team of computer scientists.
So far, Goggins and his team have collected more than 400 million tweets on cultural topics related to math, learning and breaking news.
“The ultimate goal is to be able to understand what are the categories of behavior in social media in technology—how do organizations emerge?”
“We take what we’ve learned to find ways of probabilistically estimating the likelihood of the behavior of people.”
Goggins says of TwitterGoggles, which originally went by the name TwitterZombie, “We take what we’ve learned to find ways of probabilistically estimating the likelihood of the behavior of people.”
But unlike most studies in the “big data” trend, Goggins and his team are adding the value of social science to their analyses.
“There are people out there who are taking computational predictions,” Goggins says. “With our approach, we bring actual qualitative research and manually look at what signals predict when something is changing.”
So far, Goggins and his team have collected more than 400 million tweets on cultural topics related to math, learning, and breaking news.
“For instance, when the Pope resigned, I started a job that collected any tweet using the word ‘Pope.’ Now that we have that data permanently, we’ll be able to analyze it,” Goggins says.
To do so, Goggins takes a sample of tweets to study under a social lens—either randomly, or by grouping tweets into categories, depending on the given topic—and analyzes their linguistic construction.
“There’s always a group of people who are most prevalent in a topic of tweets,” Goggins says. “So we can look at who those people are that are dominating the conversation, as well as people who are retweeted most in the conversation, and that gives us insight on who people are most likely to share.”
Goggins adds that retweeting is even more popular than you might think. He says that out of 10 million people tweeting, they’re often sharing the same 10 to 15 links.
“But there’s also a lot of discourse, and particularly in disaster, there is a lot of information sharing,” Goggins says. “Sometimes people are sharing their opinions or their affiliations.”
Another element that Goggins looks for when studying tweets are hashtags, a tag marked by the pound symbol embedded within tweets. He says his team is studying what the location of a given hashtag within a tweet might say about the tweet’s intended purpose—information that can’t be gathered by machine analysis alone.
“There is a lot of noise about big data, and I think it’s often misunderstood,” Goggins says. “But we’re not getting interesting insights from computational data—the interesting insights are coming from social scientists.”