SNU’s Yong-chan Park Named Runner-Up for 2026 SIGKDD Dissertation Award, First Ph.D. Graduate from a Korean University to Receive the Honor

- First researcher from a Korean university to receive the award since its establishment in 2008

- International recognition for the originality, scientific contribution, and technical depth of his doctoral research as a whole

- FFT-based data analysis technologies achieve up to 19× faster processing while maintaining the same accuracy

▲ Yong-chan Park, Ph.D. (right), poses for a photo after being named Runner-Up for the 2026 SIGKDD Dissertation Award.

Seoul National University College of Engineering announced that Yong-chan Park, Ph.D., who received his doctoral degree from the Department of Computer Science and Engineering in 2026 under the supervision of Professor U Kang, has been named a Runner-Up for the 2026 SIGKDD Dissertation Award, presented by ACM SIGKDD. KDD is one of the leading international conferences in the fields of data science and artificial intelligence.

Park is the first researcher to earn a Ph.D. from a Korean university and receive this award. Established in 2008, the SIGKDD Dissertation Award annually recognizes three outstanding doctoral dissertations in data science, data mining, and knowledge discovery. Candidates are comprehensively evaluated on the originality of the dissertation as a whole, the significance of its scientific contributions, its technical depth and soundness, and the quality and completeness of its presentation. The recognition is therefore particularly significant in that it acknowledges not just a single research paper, but the academic value and overall quality of the body of research Park accumulated throughout his doctoral studies on the international stage.

Park’s award-winning dissertation is titled “Fast Fourier Transform for Data Mining: Theory and Algorithms.” The dissertation extends the Fast Fourier Transform (FFT) beyond its conventional role as a transformation operation, establishing it as a design principle encompassing computation, representation, and real-time learning for large-scale data mining.

Conventionally, even when only a small portion of the available frequency information is needed, all frequency components are computed before most of them are discarded. To eliminate such unnecessary computation, Park developed a technique that directly computes only the required frequency components while controlling approximation error, as well as a method that automatically identifies optimal configurations for multidimensional data according to data size and accuracy requirements. He further extended the research by developing techniques that rearrange tensors—high-dimensional data composed of multiple axes—into structures that are easier to compress, along with online analysis methods that rapidly update continuously incoming data to reflect the latest changes without retraining the entire model.

The four core research projects—PFT, Auto-MPFT, PuzzleTensor, and FOCAL—culminated in papers published at KDD 2021, 2024, 2025, and 2026, respectively. Experiments using a wide range of real-world and synthetic datasets showed that the proposed technologies achieved processing speeds of up to 19 times faster than existing methods while maintaining the same level of accuracy. The methods also improved data reconstruction quality, compression efficiency, and anomaly detection performance, and are expected to enable faster and more reliable analysis of large-scale, high-dimensional data—including time-series data, images, and sensor streams—using fewer computational resources.

[Contact Information]

Professor U Kang, Department of Computer Science and Engineering

Seoul National University / +82-2-880-7254 / ukang@snu.ac.kr