Ten months into the coronavirus disease (COVID-19) outbreak, we are living in an era of untact, a term widely used in Korea to signify the phenomenon in which social distancing has become a given in daily life. The same applies to the fall semester of Korea University (KU), where face-to-face communication between students have become all too rare. In an attempt to lend students a hand, KU has presented a perfect service for an era of untact – the KU Artificial Intelligence (AI) Adviser, a customized course recommendation service for KU students.

KU has become the first university in the nation to develop and launch a customized AI course recommendation system for students. The unprecedented system, named the KU AI Adviser, gives students personalized recommendations for course choices based on 20 years’ worth of data on the class registration of students from the same major as the user or those who have taken similar courses. As a part of the “KU Insight Miner” research project, the AI Adviser is the first instance of KU’s attempt at a data-based AI service. The beta service of the AI Adviser has been open on the KU Portal to Information Depository (KUPID) since late July, continuing throughout the entire course registration period of this fall semester.

The purpose of the AI Adviser can be found in its Korean name, AI Seonbae. The name is coined from the Korean term seonbae, which indicates senior students, or students from an upper-grade level. “The KU Data Hub aims to help school members, including students, through the use of data,” says Ms. Lee Jin-sook, a data scientist from the KU Data Hub. “The idea for the AI Adviser stemmed from the curiosity on whether we could use data to replace the role of seonbaes,” she said. Indeed, with most school events cancelled in the wake of the COVID-19 outbreak, the AI Adviser aims to help students overcome the limited source of information regarding course choices.

Taking Advice from Algorithms

As remarkable as the idea of the AI Adviser is the mechanism behind it. “The technology behind the course recommendation system of the AI Adviser can be understood through two separate mechanisms,” explains Ms. Lee. “The first is a ‘student-based recommendation,’ which uses a user-based collaborative filtering, the traditional method for recommendation algorithms.” She further elaborated that the ‘student-based recommendation’ functions by finding a student who has given similar course ratings as the user in order to recommend the courses for which the student has given the highest ratings.

The second mechanism is a course-based recommendation, which recommends courses that are most similar to the courses the user has taken. According to Ms. Lee, the process of deciding the criteria for course similarity undergoes the following steps: the ratio is first set so that the total sum of the courses the student has taken – distinguished by major that the course belongs to – adds up to one. Then, the course registration ratio of the students who had taken the course is collected for each course. The average of this course registration ratio is seen as the ‘characteristic’ of the course. Lastly, the characteristic of each course goes under a Dimensionality Reduction through the Uniform Manifold Approximation and Projection (UMAP) algorithm, which allows courses with similar characteristics to appear closer, Ms. Lee explained.

One of a Kind

What is most remarkable about the AI Adviser is perhaps how it differentiates itself from existing services or programs. According to Ms. Lee, one thing that differentiates the KU AI Adviser is the fact that KU is the first university in the nation to provide such a service using AI. “We have been building the foundations for this service since 2019, such as by establishing a data warehouse and building up competence through data research,” she mentioned. “The AI Adviser carries great significance as we have created it by breathing life into data that has been asleep for decades and by pursuing the direction that suits our reality.”

Another aspect that makes the AI Adviser one of a kind is the development process, says Ms. Lee. “The service was developed and applied by a professional workforce within the university, with special help from the School of Industrial Management Engineering, the Department of English Language and Literature, the Department of Computer Science and Engineering, the Department of Education, the Department of Psychology, the School of Electrical Engineering, the Center for Human Ecology Research, the Academic Affairs Committee, the Center for Teaching and Learning, and the Student Counseling Center,” she said. She added that the AI Adviser was made through community activities held every other week to communicate and collaborate for the pure purpose of “student success.”

KU has also announced its plans to utilize the experience and foundations earned through the AI Adviser to advance education-related services further and develop recommendation algorithms suited for an educational environment. Regarding what one may expect from the KU Data Hub in the future, Ms. Lee mentioned that they were currently preparing to launch the second series of the KU AI Adviser – the multiple major recommendation beta service, in order to alleviate the difficulty that students go through in choosing their career paths.

The Silver Lining in Adversity

In a matter of months, the COVID-19 pandemic has put the entire world in uncharted territories. As unprecedented as current circumstances are, we have been forced to adapt ourselves with new methods, rules, and technologies. Yet adversity has a tendency to make people bond, to push them past previous limits in order to unveil new possibilities and potentials. The KU AI Adviser is a perfect example in the sense that it is a combination of the talents and hard work of those who wish to contribute to making a better world during a challenging time.

“While the KU AI Adviser – it being an unprecedented attempt – is not without its flaws, it is a product of the raw data from our very own school and not just store-bought like any other electronics,” says Ms. Lee. Though it may be a product of numbers and technologies, the KU AI Adviser seems to have indeed brought some much-needed human warmth into the lives of students in this age of untact.