本発表は，医療記録データにおける数量表現の適切な抽出のために，医学分野の用語集・辞典と医師国家試験問題文より抽出した「第」を冠する序数詞について「数詞の表記のゆれ」と「名義尺度と順序尺度のどちらにあたるのか」の2 点に注目して分析を行ったものである．調査の結果，立項される序数詞においては，同じ見出し語であっても数詞の表記法にゆれがあること，算用数字は順序尺度が多く名義尺度が少ないこと，ローマ数字は名義尺度が多いことなどを示した．ただし，序数詞の特性上，尺度の判別に迷うものも多かった．今後，表記のゆれを所与とし，医療記録データから数量表現を的確に抽出するためには，序数詞だけではなく，数詞と助数詞，および隣接する名詞との関係を明らかにする必要がある．
In this study, for appropriate extraction of quantitative expressions from medical record data, we analyzed ordinal numerals including the Japanese character “第 (dai)” extracted from glossaries and dictionaries used in the medical field and the National Medical Examination for Doctors, focusing on two points: fluctuations in the notation of numerals and whether the numerals correspond to nominal or ordinal scales. The results of the survey showed that (1) there are variations in the notation of numerals in ordinal numerals, even for the same headword; (2) arithmetic numerals are most often on ordinal scales but sometimes on nominal scales; and (3) Roman numerals are most often on nominal scales. However, due to the characteristics of ordinal numerals, there were many cases where it was difficult to determine the type of scale. In the future, it will be necessary to clarify the relationships between the scale and not only ordinal numerals but also numerals, particles, and adjacent nouns to accurately extract quantitative expressions from medical record data, given the variation in notation.