UNIVERSITY PARK, Pa. — In Penn State's College of Information Sciences and Technology (IST), many graduate students are researching the computing methods that can make social media more engaging. It’s only natural that Fei Wu, Hongjian Wang and Feng Sun — all doctoral students in the college — were sought out for their expertise with summer internships at Snapchat, Twitter and Facebook, respectively.
By working with Associate Professor of IST Jessie Li on their doctoral research, Wu and Wang have become well versed in expanding the insights drawn from analyzing large databases, a process known as data mining. Sun, whose research focuses on increasing civic engagement through the use of technology, spent the summer at Facebook, though he is not able to discuss his work publicly due to the confidentiality of his project.
For Wu and Wang, however, they are intrigued at how to decipher the meaning behind the extraordinary amount of data now available through technologies like GPS and social networks. Their end goal is to better understand how people interact with the world.
Wu said of his research at IST, “We work on bridging them together. If I have a [GPS point], I want to answer a question — what are you doing?”
In Wang’s academic research project, Li and the team of graduate students use data mining techniques to better understand factors surrounding urban environments, such as crime rates, traffic, maps and demographics.
Said Wang, “We merge all kinds of data to understand these cities better.”
They’ve also incorporated information gleaned from FourSquare, a location-based social media app that enables users to “check in” at points of interest (POI). By integrating the POI data from the app, they are able to give more context to people’s movements.
“Traditionally, to understand a region, people just looked at demographics and maps,” he explained.
But with these new data sets, the students are able to gain more insight into why people travel to certain locations, whether it’s a grocery store for shopping or an emergency room for an injury. And with every facet that’s incorporated into the IST algorithm, the researchers are hopeful they can improve people’s everyday lives by helping urban planners create more efficient and safer cities.
“Everything in a city is connected,” Wang said. “Machine learning gives us that insight.”
Understandably, that expertise was the reason they were recruited for internships at Twitter and Snapchat. Wu collaborated with a team focused on increasing the effectiveness of rankings within sponsored posts that are shown to users based on a complex algorithm.
“We are improving that ranking; if you see that user’s Story, you’re more likely to click on it,” he explained.
At Twitter, Wang worked with the prediction team. At face value, their task is easy to understand.
“Twitter makes money by showing ads,” he said. “But how do we give people the best ads [for them]? That’s machine learning.”
He likens the process to his research at IST, saying, “In essence, it is using all different kinds of data sources to predict a problem.”
The difference is when Wang helps to refine the process used at Twitter, they are able to gain a tremendous amount of revenue. Emboldened by his academic research, he suggested incorporating urban data into Twitter’s ad display algorithm, and said his supervisors were interested in the possibility.
The students also enjoyed the employee perks synonymous with tech startups — like movie pre-screenings at Snapchat, and trips to Alcatraz, go-karting, and a dedicated gaming room at Twitter. Wu also commented he liked to occasionally play around with Snapchat’s well-known filters.
“I like those ugly filters, the ones that change your face,” he said with a laugh.
With professional experience under their belts, these students know their foundation in IST positions them well for their future careers.
Said Wang, “In machine learning and data mining, they need a ton of people with that skill. Everything is about information and technology.”
Wu agrees their skillset is valuable to tech companies.
“At IST, there’s a broader perspective into real world problems,” he said. “We always take things in a computational aspect, but IST adds the social aspect. So when you think about a problem, you have this different perspective.”