Evaluation of Korean Machine Reading Comprehension Generalization Performance Based on Domain-Specific Evaluation Datasets
Domestic Conferences & Theses
Authors
Hyelin Kim*, Sanghyun Cho, Joongmin Shin, Hyuk-Chul Kwon
Abstract
Identified the limitations of existing tabular machine reading comprehension models in domain generalization through cross-validation, highlighting the importance of constructing tabular datasets across diverse domains.
Key Contribution
Identified domain generalization limitations in tabular MRC models through cross-validation analysis.