Undergraduate Students Help Improve CONNECT's Matching Process

Algorithm Features

One of the key features of the CONNECT program is its ability to quickly identify quality student-project matches with an algorithm developed in conjunction with a group of UT Austin undergraduate computer science students. The CONNECT program algorithm significantly improves the efficiency of its matching process and helping build capacity within the program itself as it continues to expand.  

In recent semesters, the CONNECT program has drawn immense student interest, receiving anywhere from 40-60  applications from interested students each fall, spring and summer cohort -- about three times as much interest as there are projects available. Prior to developing the algorithm, the CONNECT program’s manual matching process would usually take a team of three or four people several hours to identify quality matches. 

“Our goal was to speed up the process of matching and reduce our own bias from the matching process,” said Ethan Tenison, Project Manager for Data Initiatives at the RGK Center.  

In spring 2020, the CONNECT team partnered with a group of undergraduate students from the Cockrell School of Engineering engaging in an honors capstone course through the Electrical and Computer Engineering Department. The goal of the collaboration was to develop a scalable system that would automate the matching process between graduate students and organizations in the CONNECT program. By the end of the semester, the team developed a website that can be used by students, organizations, and CONNECT staff to make the process more efficient and accurate for all.  

“Not only did the algorithm they developed enhance efficiencies for our matching process, but the questions asked by the student group along the way provoked a lot of additional reflection on how to improve other processes related to student and project information intake, too,” said Alyssa Studer, CONNECT Senior Student Program Coordinator.  

Now, the CONNECT team uses this “machine-guided” process of quickly matching community organizations with a data challenge to graduate students with skills in data management and analysis. After plugging in the information from applications into the algorithm, the team reviews the suggested matches and accounts for any special considerations or recommendations. The entire process takes about an hour.  

“Using the matching algorithm significantly reduced the time it takes to pair students with organizations, and it helps us create a fairer matching process,” said Tenison.  

The algorithm the team developed is a variation of the Gale-Shapley algorithm, which is often used for dating websites and for matching people who need kidneys with suitable donors. CONNECT staff can input into the algorithm information from student and organization applications and receive suggested matches in about five seconds, along with a description of why the algorithm found that match to be ideal.  

A screenshot from the team's final project video of the undergraduate students who worked with CONNECT on the matching algorithm. 

ThienSon Ho, one of the undergraduate students involved in the project, developed the user interface and front end of the website, which includes the main pages, surveys, and the administrator dashboard. He also worked on developing the database interactions and the server used by the algorithm.  

“The most rewarding part was knowing our project was going to help real students and community organizations,” said Ho. “I’m glad we were able to create something to help fellow UT students, our local community, and the amazing CONNECT team.” 

 

Special thanks to Arman Khondker, Desiree Tang, Josh Papermaster, Punit Patel, and Thienson Ho for their work on the CONNECT matching algorithm website. Thank you to Professor Christine Julien, Ph.D. the student group’s faculty member and faculty member in the Department of Electrical and Computer Engineering. The photos in this article are stills from the team's final project video

Aug. 9, 2021