Tips from an “AI Mentor” for new semester’s course registration
KU is the first university in Korea to offer customized recommendations for liberal arts courses
An innovative service in the era of “untact” when face-to-face interaction is difficult

Following its recent customized and personalized service system for students based on data analysis/management/feedback, Korea University (President Chung Jin Taek) has released another innovative service for training people-centered and future-oriented personnel.

KU is the first university in Korea to develop and implement an “AI Mentor” program in which an AI system recommends liberal arts courses to suit the interest of individual students.

The system gained the name, “AI Mentor” because its recommendations are based on actual cases. It first investigates the past 20 years of data on students from the same department or those who have taken similar courses, and then recommends preferable courses to individual students.

It is especially noteworthy that this customized service offers automated suggestions in today’s era of “untact,” when students cannot meet other students to receive in-person advice due to COVID-19.

In order to increase the recommendation performance and satisfaction of the AI Mentor program, various hyper parameters have been employed based on a “subject similarity algorithm” developed by KU’s Office of Digital Information and the “collaborative filtering algorithm,” Netflix’s renowned movie recommendation algorithm.

Accordingly, three recommendation models have been developed, and each model suggests 21 courses, each different from the other. The KU Office of Digital Information collected students’ opinions on the three models, adjusted the hyper parameters several times to improve the performance, and finally selected and implemented two models after a process of re-learning and verification. Along with the list of course names, the system also shows a visualized responsive diagram that demonstrates the core competencies, the proportion of students in their first major and second major, and a distribution of students based on academic year.

It takes as little as 16 hours and up to 35 for the AI Mentor to learn related data before it extracts, analyzes, and offers optimal customized coaching to the students. The system is the first instance of data-based AI service developed as a part of the “KU Insight Miner” R&D Project under the “University Innovation Support Project” supported by the Korean Ministry of Education. Based on the opinions on the AI Mentor for student success, KU plans to improve the learning and service infrastructure for machine learning and deep learning, and continue to develop algorithms suitable for the education field.

The AI Mentor will be available via the campus portal from July 21, 2020 to mid-September, when the course add/drop period for the second semester ends. The system is expected to serve as a guide for students before they register for the new semester.

In addition, KU established a so-called Data Warehouse (Data Hub) in 2019, and is operating a data analysis community. This community consists of professors, students, data scientists, data engineers, and programmers from various departments including the Department of Education and the Department of Psychology, and various organizations including the Office of Academic Affairs and Center for Teaching and Learning. Professors Han Sung-won (School of Industrial Management Engineering) and Nam Ho-sung (Department of English Language and Literature), participating in the development of the AI Mentor service, are a member of the data analysis community. The data analysis community is analyzing its big data to define student success, to look for convergence among majors, and to develop a diversity index.

On August 11, 2020, the KU Center for University Innovation Support will host the 3rd KU Innovation Workshop to discuss the experiences and efforts for student success in the post-corona era, such as the AI-based data analysis system.