Korean business documents frequently contain mixed numerals, Chinese characters, and Latin letters; this multi-script mixing increases the difficulty of character recognition and is one of the main reasons why general-purpose OCR engines perform inconsistently in Korean scenarios.