The research team led by Professor Sangki Ko from the Department of Artificial Intelligence at the University of Seoul has achieved a significant milestone with their paper “RESYN: A Generalized Recursive Regular Expression Synthesis Framework” being accepted at the esteemed IJCAI 2026 conference. This study tackles the “Programming by Example (PBE)” challenge by introducing an innovative recursive AI framework named “RESYN,” designed to generate complex regular expressions from examples efficiently. Unlike previous studies that used simpler benchmark data, this research delves into the intricate structural complexity of real-world regular expressions. The team developed a compact neural network model, “SET2REGEX,” which employs a divide-and-conquer approach to problem-solving and emphasizes permutation invariance among input examples. This method outperformed existing large-scale models and general-purpose language models on complex datasets, achieving high performance with fewer parameters. The paper’s co-authors include Professor Sangki Ko, Master’s student Sungmin Kim from the University of Seoul, and researcher Yoseop Han from Yonsei University. The team anticipates that their findings will extend beyond regular expression synthesis to various program synthesis tasks and complex AI problem-solving.
Has Professor Sangki Ko’s Team at University of Seoul Achieved Recognition at IJCAI 2026?
