Faces of UMSI: Cheng Li

UMSI doctoral student Cheng Li says her research is driven by her desire to develop interactive machine learning and data algorithms that can help people.

“I want to be able to help people who aren’t experts in data mining or machine learning, but they need to solve particular tasks in their work,” Cheng says. “I'm thinking about developing systems that could help people understand the learning process of the algorithms, and interact with people, so they can utilize tools and adapt those tools to their particular domain.”

Cheng’s interest in machine learning, data mining, and information retrieval grew while she was working toward her bachelor’s degree in computer science at China’s Zhejiang University. While studying at Zhejiang, Cheng spent some time as an exchange student and research assistant at Singapore Management University, where her advisor worked on a number of projects that involved data mining. 

As she began to read more papers on the topic, she came across several articles published by UMSI Associate Professor Qiaozhu Mei and was inspired by his work. After learning more about Mei’s research and the opportunities available at UMSI, Cheng decided it would be the best place for her to pursue her PhD and work toward her lifelong goal of becoming a professor.

“My father is a biology professor back in China, so it was my dream to become a professor too,” Cheng says. “UMSI is the only information school I applied to because I am really interested in the interdisciplinary atmosphere here, and everyone here really cares about the connection between people and technology.”

Cheng is a member of Mei’s Foreseer research group, which finds broad applications in web search and mining, social computing, scientific literature mining, and health informatics by focusing on novel methods and applications of information retrieval, text mining, natural language processing, and social network analysis. With the group, Cheng is currently developing an interactive retrieval system that can help users access all the relevant results related to their queries. 

While people who use popular Internet search engines are typically only interested in the top one or two results, Cheng says the system she is working on would be particularly beneficial to users doing literature surveys who would need access to all the papers related to their topic, or physicians who need to review all the patients with certain conditions for clinical trials, or attorneys trying to find every piece of evidence related to their case from documents that are under legal hold.

The system would ask a user to judge documents and determine if they are related to the query and would propose new queries that allow users to edit and help the system to retrieve more relevant documents. 

Cheng is also exploring how machine learning and data mining can be applied to issues in social media. In the summer of 2014, she worked as a research intern at Twitter and recently published a paper on her research that looked to improve the accuracy of predicting whether people will click on ads.

The paper, "Click-through Prediction for Advertising in Twitter Timeline," was accepted by the 21st ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD), a premier conference that brings together researchers and practitioners from data mining, knowledge discovery, data analytics, and big data. 

Cheng will go to Australia in August to present her research at the conference, which accepted her paper in its Industry & Government Track. Submissions to this category focus on systems that address real-world problems, something that Cheng is happy she’s been able to do in her time at UMSI as well.

“I’m most interested in machine learning and data mining algorithms and how to apply them to real problems to help people,” Cheng says. “And that’s what we do here—we look at real problems and we solve those problems for people.”