Earticle

초록

영어

This study explores a Shakespeare Research Article Abstract Corpus through topic modeling, a machine-learning technique that automatically identifies topics in a corpus. First, we identify which topics are prominent through the entire corpus. We also investigate the top 20 topics in each particular decade such as the 1980s, 1990s, 2000s, and 2010s and examine patterns, trends and ranking changes such as falling, rising, and curve contours over time. In addition, we extract corpus keywords using the cross-validation method on Wordsmith tools 6.0. in order to compare similarities and differences between topic modeling keywords and corpus keywords. We also select each group of absolute keywords which have zero frequency in reference corpora to examine which words are associated with new trends in each period and to explore which shared common words are found in topic modeling and corpus keywords. Finally each group of non-absolute keywords extracted from the three corpora is discussed to check patterns and trends identical to topic modeling. The results of this comparison conform that it is hard to assert that topic modeling keywords are well grouped into certain research subjects over corpus keywords and show better trends over time than corpus keywords. This is because both topic modeling keywords and corpus keywords show their own respective merits.

목차

Abstract
 1. 서론
 2. 선행 연구
 3. 연구자료 및 연구방법
  3.1. 자료통계
  3.2. 연구방법
 4. 연구결과
  4.1. 토픽모델링 결과 분석
  4.2. 토픽변화 양상과 추이유형
  4.3. 토픽모델링 키워드와 코퍼스 키워드 간의 비교
 5. 결론
 참고문헌

저자

• 장세은 [ Se-Eun Jhang | 한국해양대학교 ] 제1저자
• 이수상 [ Soo-Sang Lee | 부산대학교 ]
• 송원문 [ Won-Moon Song | 신라대학교 ]
• 정해룡 [ Hae Ryong Jung | 부경대학교 ]
• 이성민 [ Sung-Min Lee | 한국해양대학교 ]
• 김재훈 [ Jae-Hoon Kim | 한국해양대학교 ] 교신저자

참고문헌