Analyzing the data that drives media and music consumption

Courtney Allen
October 01, 2018

UNIVERSITY PARK, Pa. — As the information age gives rise to new technology, it also is revolutionizing two of the most longstanding and traditional industries: news media and music. Thanks to internships with The Washington Post and Spotify, two doctoral students in the College of Information Sciences and Technology are at the forefront of this digital revolution.

Examining the data behind the newspaper

Doctoral student Yiming Liao spent his summer focusing on generating news timelines for stories published by The Washington Post as well as user engagement, using data analysis.

Prior to starting his position, Liao was given the opportunity to speak with the director of the data sciences team at The Washington Post, which solidified his interest in the company. 

Yiming Liao

Yiming Liao, a doctoral student in the College of Information Sciences and Technology, spent his summer as a data sciences intern with The Washington Post, where he built prediction models that helped the newspaper better understand which types of content are most interesting to readers.

IMAGE: Provided

“Having chatted with the director as well as the data scientists on the team, I found that they are very nice and open in terms of what I am interested in and what I want to do in this particular field,” said Liao. “So I was in.”

Liao’s role at The Washington Post was to leverage data-driven approaches to help the newsroom. Specifically, he worked on news timeline summarization, which automatically generates dated summaries for news events, such as the North Korea summit or the U.S.-China trade war.

His research helps journalists create these timelines, because generally a news event can range from months to years and involve thousands of news articles. The time and effort expended by the journalists can be saved with an auto-generation tool.

“My internship is more like problem-driven and exploring on different approaches, so it is hard to describe a day-to-day task,” said Liao.

As part of his internship duties, Liao built prediction models that helped the daily newspaper better understand which type of stories and content would be the most interesting to their readers. According to Liao, the process begins with a problem/issue that is of interest to The Washington Post.

With the help of journalists throughout the company, Liao explained, he and the data sciences team processed and analyzed the discussion and input of the journalist, to either improve existing methods for solutions or adapt to them for higher rates of efficiency.  At the end, his team would produce a new prediction model for The Washington Post to build from.

“Through literature reviews, we would determine intended approaches, adapt them into our system, improve existed methods, and evaluate the solutions,” said Liao “Finally, we will deploy them to production and do iteration with feedbacks. It is very similar to conducting research at labs.”

Liao shared that the education he’s received from the College of IST prepared him with the necessary skills for tackling the complex projects at his internship. Reviewing past research and formulating new questions are two major components of his daily responsibilities at The Washington Post, and are also key aspects in the IST courses he takes.

Understanding mass user music consumption

Spotify, one of the world’s leading streaming music services, founded by the Swedish company Spotify Technology, is known for its vast music library. According to College of IST doctoral student Lu Liu, who served as a data scientist this past summer, innovative technology is a key part of the company's success.

Lu Liu

College of Information Sciences and Technology doctoral student Lu Liu completed a summer internship as a data scientist with Spotify, where her research team focused on understanding user consumption behavior from mass streaming data. 

IMAGE: Provided

In her internship, Liu focused on understanding user engagement and analyzed the effectiveness of the company’s different approaches to increase these metrics. When a user listens to a song, she explained, Spotify can recommend new music based on past listening behaviors. The more music someone listens to, the more the service learns about what they like and recommends future songs that meet their needs.

“Our research team focuses on understanding user consumption behavior from mass-streaming data,” said Liu, who chose the internship because of its focus on data analysis. “We would like to increase customers’ engagement by providing better music recommendation services.”

Each team at Spotify is composed of different people from various departments, which allows for a variety of perspectives when problems occur. Liu explained there are tech researchers, user researchers, and product and content creators that collaborate and use data science methods to analyze and improve user engagement, something she experiences in her research at IST.

“IST truly provided me with experience of working and communicating with people from diverse backgrounds,” said Liu.

According to Liu, Spotify’s work environment is “open and free flowing,” connecting her with people outside of her field. This environment enabled her to utilize the networking skills she had developed working on group projects in the College of IST.

“[My bosses] encouraged interns to communicate more with employees, especially with the ones not even related to your project,” said Liu. “Communication is always important for better ideas.”

Liu explained how her internship at Spotify and experience in IST connects with her interest in the music industry. She desires to keep forging paths that allow her to intertwine both music and her research in IST.

She encourages College of IST students and alumni seeking new career opportunities to be open to other industries that they have interest in to find a way to bridge their tech passions.

“I would suggest finding a good company culture as well,” she said. “It’s important when seeking a job. Then get connected, find someone that is similar to your interest, and contact them about your desires.”

Last Updated October 02, 2018