header

Graduate student team from the Department of Computer Engineering wins the Minister of Science and ICT Award in the K-Digital Challenge by developing an AI-based education content creation service to improve accessibility in learning

The “KHU-reka” team, composed of graduate students from the Department of Computer Engineering, won the Grand Prize (Minister of Science and ICT Award) at the K-Digital Challenge: NET Challenge Camp Season 12, hosted by the National Information Society Agency. This competition utilizes KOREN (Korea Advanced Research Network)—established to verify the development and results of application services and related technologies for high-speed networks—to develop innovative ideas in the field of Artificial Intelligence (AI) and network convergence while exploring their potential for commercialization. Ten teams competed in the final round of the competition.

Generating intuitive visualization materials for complex mathematics and science subjects

Utilizing the high-speed network and High Performance Computing (HPC) resources of KOREN, the team developed “VizClass: Real-Time AI Educational Content Audio-Visual Interpretation Service.” This service automatically analyzes and visualizes the educational content requested by the user.

VizClass was designed to narrow the education gap caused by regional and income disparities and to lower the barrier to entry in the science and engineering fields. Haeun Woo, the team leader, explained the motivation behind the service: “The gap in accessibility to educational resources between metropolitan and non-metropolitan areas persists, and the disparity in private education expenses according to income levels is increasing. Mathematics and science, in particular, involve complex and difficult concepts, leading to a significant gap depending on the learning environment.”

VizClass is built upon three core technologies: customized commentary video generation, parallel Script2Video technology based on visualization points, and real-time video data streaming via the KOREN network. When a learner uploads educational materials, AI analyzes them to generate a commentary script and extracts visualization elements for each section to automatically produce an explanatory video. The entire process is handled through a real-time, automated system implemented as a web application.

Data analyzed by AI is transmitted via the KOREN network to a high-performance computing center, where it is processed by a parallel computing-based video generation module. Once complete, the video is organized sequentially using controller-based scheduling technology and provided to the user through real-time streaming. This approach has more than doubled the processing speed and simultaneous performance compared to traditional sequential processing methods. Team member Jiseung Bae noted, “Thanks to KOREN, we were able to conduct development in a stable and fast network environment. We hope that in the future, students in underserved areas will also be able to access lag-free video content through the high-speed KOREN network.”

Expanding the AI commentary ecosystem from textbook explanations to academic papers

Existing education AI technologies have primarily focused on problem recommendations or text-based explanations. In contrast, VizClass utilizes AI to analyze learning materials, automatically extract visualization elements, and generate educational videos based on that data. Team member Sumin Son stated, “Beyond the basic concepts found in current textbooks, we plan to expand into an AI commentary ecosystem by supporting the visualization of advanced learning materials, such as academic papers, in the future.”

The performance of VizClass was verified through practical testing by comparing a standard internet network with the KOREN network under identical conditions. The results showed that processing time for a single request was reduced by approximately 45.3% compared to the standard network, and generation time improved alongside video length. The number of simultaneous connections also increased on KOREN, enabling stable service even in high-quality video streaming environments.

The service development process resulted in three research papers presented at international and domestic academic conferences, and the technologies related to VizClass have been officially registered as software with the Korea Copyright Commission. Additionally, the team received the SK Broadband Award, a special category prize. As a benefit of this award, team members will receive preferential screening and bonus points when applying for employment at SK Broadband.