Rouslan Karimov: Analyzing Networks, Online and Off

Rouslan Karimov in the Tech + Narrative Lab

Diane Baldwin/RAND Corporation

Within the first few weeks of his arrival at Pardee RAND, Rouslan Karimov (cohort '15) attended a student seminar that changed the course of his academic career.

"Andrew Cady, who was a second-year student at the time, gave a talk on the use of Twitter data in research at RAND," he said. "I was fascinated. I'm not sure why I liked it, but it piqued my interest, and I talked with him about it after his seminar."

Soon after, he attended a talk by Prof. William Marcellino on conducting network analyses of tweets about ISIS. "That was probably my first introduction to network analysis," Karimov said. "It really intrigued me."

As the icing on the cake, Karimov's first year advisor, Prof. Sarah Nowak, introduced Karimov to Luke Matthews, because she knew they shared an interest in cultural networks. Matthews, now the codirector of the Center for Applied Network Analysis, would go on to become Karimov's mentor, coauthor, and dissertation committee chair.

Karimov explained that network analysis "was a surprising and exciting discovery for me, since it offered me a completely different way of looking at certain phenomena that I had studied before."

Karimov came to Pardee RAND with more than ten years of experience as a statistics and monitoring specialist, but while the work involved networks, it hadn't involved network analysis.

"I worked on vaccination statistics at UNICEF and witnessed firsthand the negative impact that vaccine-related conspiracy theories can have in different countries," he said. "Understanding how such theories spread and how to counter them is hugely important, especially considering the recent developments in election meddling, government-sponsored influence operations and global paranoia triggered by the coronavirus pandemic."

“Network analysis offered me a completely different way of looking at certain phenomena that I had studied before.”

His academic background includes an undergraduate degree in economics as well as a master of public administration degree with a focus on economics.

That experience has been useful, he said, as network analysis is also quite quantitative. "Some network analysis techniques are direct modifications to methodologies used in econometrics, for example," Karimov explained.

He eagerly took the Social Network Analysis course when it was offered his second year. Around the same time, Cady stepped down from his role as the Twitter data coordinator, a position managed by RAND's Information Services that allowed numerous research projects interested in social media and social network analysis to query and download historical data from the Twitter API. Remembering their talk, Cady asked if Karimov would be interested in taking over. Karimov didn't hesitate.

"Many of my projects came about because of my serving as the RAND Twitter data coordinator for several years," he said. "These data are perfect for network analysis; in fact, it is required if you are examining them with any kind of rigor. While I did not perform such analyses for all Twitter-related projects, I definitely did for some."

Because they are often conflated, he is quick to point out that social media analysis is only one type of network analysis.

"Network analysis accounts for interconnectedness in general, and it models systems where actors are not necessarily independent. I am very interested in the spread of information through communication networks, but social media is just one example," he said.

Karimov conducted a network analysis of dark web sellers during a fall 2019 hackathon. He presented his findings virtually.

Todd Richmond/RAND Corporation

Karimov says about a third of his project work has involved network analysis in some way, largely because of his role as Twitter data coordinator, but he has also used the method in some extracurricular areas.

Last fall, he participated in a Pardee RAND hackathon that had students explore two years of illicit market data from RAND's Dark Web Observatory. "One of the things I did was try to trace linkages of sellers in different marketplaces," he said.

Not surprisingly, network analysis methodology is also central to Karimov's dissertation. In fact, it grew out of the final paper he wrote for his Social Network Analysis class. "It started as an assignment to find data amenable to network analysis," he said, and "ended up with me realizing I could mine the topic enough to write a thesis about it."

His working title is "Visa-free travel policies: determinants and effects" and in it he is mining eight years of data to explore how visa-free privileges both influence, and are influenced by, tourism, migration, and bilateral trade and investment flows between countries.

"All of these relationships form global networks which require special statistical techniques for proper analysis," he said.

Karimov expects that one of the "major contributions of my dissertation should be correcting previous research that did not take these network dependencies into account."

As far as the future holds, he would like to continue to use his network analysis skills, but he knows they may not be accepted or understood everywhere. "I liked what I was doing at the UN and would love to go back, but network analysis wasn't used much there," he said. "I'm also looking at places like Graphika, which would be a completely different direction, but I have diverse interests. I guess we'll see!"

— Monica Hertzman