KPoEM dataset

KPoEM (Korean Poetry Emotion Mapping) Dataset We collected 483 poems from five prominent Korean modern poets: Kim So-wol (165 poems), Yun Dong-ju (112 poems), Im Hwa (43 poems), Yi Sang (46 poems), and Han Yong-un (117 poems). We scraped their major works from public domain sources, including "Azaleas" (Korean: 진달래꽃, Jindallaekkot), "Sky, Wind, Stars, and Poetry" (Korean: 하늘과 바람과 별과 시, Haneulgwa baramgwa byeolgwa si), "Hyunhaetan" (Korean: 현해탄, Hyeonhaetan; Hanja: 玄海灘) , "Silence of My Beloved" (Korean: 님의 침묵, Nimui chimmuk), In the case of Yi Sang, the dataset focuses on his Korean-language series poems such as "Crow's Eye View" (Korean: 오감도, Ogamdo; Hanja: 烏瞰圖), "Reverse" (Korean: 역단, Yeokdan; Hanja: 易斷), and "Critical Condition" (Korean: 위독, Widok; Hanja: 危篤). In results, the dataset includes a total of 7,622 emotion-annotated units: 7,007 line-level annotations and 615 work-level annotations. Revision: In v4, the names of some columns that were originally in Korean were changed to English. Please refer to the following presentation: LIM, I., Ji, H., & Kim, B. (2025). 한국 근현대시 감정 라벨링 데이터셋 구축: 문학 텍스트의 컴퓨터 기반 감정 분류와 생성형 AI 활용을 위한 기초 연구. 제2회 한국현대문학자대회 (KorLitConf), Seoul, Korea. Zenodo. https://doi.org/10.5281/zenodo.15055795 All source code used in this study has been made publicly available in the following repository. See the link for details. https://github.com/AKS-DHLAB/KPoEM