Abstract
Introduction: By automating to provide furigana for all kanji and English for disease names, it may ease the creation of teaching materials for economic partnership agreement (EPA) nurse candidates.
Method: Using morphological analysis, we affixed furigana on all kanji of the questions in the national nursing exam. These furigana were then compared to actual questions for EPA candidates. We make an additional dictionary based on the errors, and reevaluated. We examined the names of diseases and foreign names appeared in the exam.
Results: Morphological analysis using the default dictionary showed that the rate of errors was about 9%. Additional dictionary were made from the words of errors, and using the additional dictionaries showed that the rate of errors improved to 1-2%. There were 722 terms for the names of diseases and 22 for the names of foreigners.
Discussion: We showed it is possible to automate affixing furigana and English to the questions of the national nursing exam. Despite limits to automation, it is possible to reduce human error.
