한국어 일반 명사의 기능적 분포와 분류
Classifying Korean Common Nouns by Their Distributional Similarities
2
조회
0
다운로드
2
인용
0
참고문헌
키워드
초록
This paper attempts to classify Korean common nouns based on the distributional hypothesis. In the Korean PoS (part-of-speech) system, common nouns are no longer being classified separately. However, this research demonstrates that the usage patterns of common nouns can still be distinguished into several types, despite the lack of formal classification. To achieve the classification of Korean common nouns, a 2-gram vocabulary set was created from the Korean lexical semantic corpus. From this set, items containing common nouns were processed separately to construct the analytical data for this research. The 'common nouns (1566) x 2-gram items (531)' table containing frequencies was subjected to Positive Pointwise Mutual Information (PPMI) and Principal Component Analysis (PCA) clustering. The results confirmed that Korean common nouns can be divided into five or more categories based on this analysis. The significance of this research lies in demonstrating the ability to classify common Korean nouns based on n-gram distributions only using unsupervised learning methods.
Abstract
This paper attempts to classify Korean common nouns based on the distributional hypothesis. In the Korean PoS (part-of-speech) system, common nouns are no longer being classified separately. However, this research demonstrates that the usage patterns of common nouns can still be distinguished into several types, despite the lack of formal classification. To achieve the classification of Korean common nouns, a 2-gram vocabulary set was created from the Korean lexical semantic corpus. From this set, items containing common nouns were processed separately to construct the analytical data for this research. The 'common nouns (1566) x 2-gram items (531)' table containing frequencies was subjected to Positive Pointwise Mutual Information (PPMI) and Principal Component Analysis (PCA) clustering. The results confirmed that Korean common nouns can be divided into five or more categories based on this analysis. The significance of this research lies in demonstrating the ability to classify common Korean nouns based on n-gram distributions only using unsupervised learning methods.
참고문헌 (17)
- 김정남, "조사 없이 사용되는 한국어 명사류에 대하여― 한국어 교육 내용에 대한 제안을 겸하여", 어문론집, 72, pp.7~33 (2017), 중앙어문학회[DOI]
- 김정남, "한국어 체언과 조사의 결합 제약에 대하여", 國語學, (61), pp.149~180 (2011), 국어학회[DOI]
- 목정수, "한국어 관형사와 형용사 범주에 대한 연구 ― 체계적 품사론을 위하여 ―", 언어학, (31), pp.71~100 (2002), 사단법인 한국언어학회
- 김정남, "韓國語 語彙 敎育을 위한 用言의 制約 活用形과 特殊 活用形 硏究", 진단학보, (109), pp.345~368 (2010), 진단학회
- 오민석, "일부 ‘수식언’의 품사에 대하여 -‘맨’, ‘한’, ‘약(約)’, ‘총(總)’을 대상으로 -", 언어와언어학, (41), pp.105~126 (2008), 언어연구소
- 이도길, "어휘의미 분석 말뭉치 구축" (2020), 국립국어원
- 이선웅, "국어의 한자어 '관형명사'에 대하여", 한국문화, (26), pp.35~58 (2000), 규장각한국학연구원
- 이선웅, "韓國語 名詞類語의 名詞性 檢證", 어문연구(語文硏究), 42(1), pp.37~62 (2014), 한국어문교육연구회[DOI]
- 이영제, "명사류의 명사성 판단 규준과 어근적 단어", 國語學, (72), pp.289~321 (2014), 국어학회[DOI]
- 이영제, "비전형적 명사류 연구의 현황과 쟁점", 國語學, (99), pp.511~557 (2021), 국어학회[DOI]
- 최운호, "'관형절+일반명사' 구성의 접속 기능에 대한 계량적 연구―'가운데'와 '경우'를 중심으로―", 한글, (269), pp.135~160 (2005), 한글학회
- 최운호, "레벤시타인 거리를 이용한 방언거리의 계산과 몽골어의 분석", 알타이학보, (20), pp.127~160 (2010), 한국알타이학회[DOI]
- 최운호, "한국어 ‘용언 어간 + 어미’ 결합의 양상 - 용언별 결합 어미 분포의 집중화 경향에 대하여 -", 언어학, (71), pp.17~47 (2015), 사단법인 한국언어학회[DOI]
- 최운호, "한국어 '명사 + 격조사' 결합의 양상", 언어학, (78), pp.3~30 (2017), 사단법인 한국언어학회[DOI]
- Goldsmith, John, "Learning phonological categories", Language, 85, pp.4~38 (2008)
- Greenberg, Joseph H., "The measurement of linguistic diversity", Language, 32(1), pp.109~115 (1956)
- Mayer, Connor, "An algorithm for learning phonological classes from distributional similarity", Phonology, 37(1), pp.91~131 (2020)