During her junior year, computer science major Emily Gao ’24 was casting around for senior thesis ideas, something that could have real-life impact, when the undergraduate course registration process sprang to mind.

While it’s not “a life-changing issue,” it can be frustrating for students to get just one, or none, of the courses they want, says Gao, who wrote an honors thesis and is now working for Boston Consulting Group after graduating in June. “It came to me that this could be a very cool problem to study because it would have implications for my life and other students in the future.”

As it turned out, Gao didn’t just study the issue, she used her algorithmic understanding to help create and implement a solution. 

She worked on the problem with her thesis adviser, Deeparnab “DeepC” Chakrabarty, a professor in the computer science department, over the course of her senior year. They analyzed anonymized information provided by the Office of the Registrar and built what would later be dubbed the “fairness algorithm,” designed to enable the greatest possible number of students to enroll in their top choices during course registration before the start of each term. 

Gao, who had taken an advanced algorithms course with Chakrabarty, says she considered the project interesting “from a CS theory point of view” but never expected it to be put into practice.

Yet after hearing from the registrar’s office about her thesis, the staff at Information, Technology and Consulting were “immediately intrigued,” says Thomas Joseph, an enterprise software architect. 

“After our first meeting with Emily, it became clear that the implementation of their fairness algorithm could significantly enhance the registration process,” Joseph says. “In partnership with the undergraduate registrar and Emily, ITC was eager to explore and develop a technical solution to improve the student election experience.”

During the months following graduation, Gao and Joseph worked to transform her thesis into a practical application. It was no small task, given the many factors—such as class year, major, and minor—that can boost the priority level for a student signing up for a course. It’s made all the more complicated because those rankings can vary from course to course. 

Gao, ITC, and the registrar’s office used “rigorous testing and quality assurance processes” to ensure the functionality and effectiveness of the algorithm, says Joseph, who, like his colleagues, was “deeply impressed” with Gao’s dedication and hard work. 

While Dartmouth’s course election process had previously rivaled or surpassed those of its peers in terms of fairness, says Andrew Ager, senior associate registrar for research, Dartmouth officials welcomed the improvement.

“We were doing really well, and Emily found a way to make it even better,” Ager says. Her approach, drawing upon her intensive study of a subject of interest to create an important upgrade, “is well aligned with the academic mission of Dartmouth.”

During the course election process, the new algorithm identifies students who are poised to get all of the courses they want and those who might get none or only one and makes swaps between the two groups, resulting in only a slight decrease in the first group, says Ager. 

“Working with Emily on improving our course election processes has been a wonderful and exciting project,” says Registrar Eric Parsons. “By incorporating her algorithm into Dartmouth’s processes, we were able to reduce the number of students who were not enrolled in any of their selected courses during course election by 85%, to fewer than 10 students for the spring term.”

The algorithm also dramatically reduced the number of students who might otherwise have only gotten just one of their course selections.

Chakrabarty, a theoretical computer scientist, says it was refreshing to see a student’s thesis that breaks new ground become a product, rather than a scientific publication, as is more often the case in his field. 

The idea of matching courses to students is a problem in graph theory that has been well studied, but Gao’s thesis gave it a new twist, applying the notion of fairness and using techniques from undergraduate-level algorithms to improve the outcomes, he says. 

And her desire to solve the problem was heartening, Chakrabarty says. “I often tell people that if you have that, then the problem will get solved, you will work on it.”

ITC has initiated a patent application with the Dartmouth Technology Transfer Office, with Gao and Joseph as inventors. 

Gao says she is grateful to Chakrabarty and to ITC and the registrar’s office for providing her with guidance and resources to complete the project.

“If they hadn’t recognized how useful it could be for future students, I wouldn’t have even thought that I could actually implement this algorithm in real life,” she says. “I’m hoping that the students are happy with the outcomes.”