초록

용어보기논문에서 용어와 풀이말을 자동 추출한 결과로, 시범 서비스 중입니다. AI-Helper

Abstract ▼ AI-Helper

Recent studies in Korean sentiment analysis mostly focus on binary classification such as 'positive' or 'negative'. However, there are few researches have discussed on emotion analysis which involves classifying opinions into multiple categories. In this paper, we have developed machine learning mod...

Recent studies in Korean sentiment analysis mostly focus on binary classification such as 'positive' or 'negative'. However, there are few researches have discussed on emotion analysis which involves classifying opinions into multiple categories. In this paper, we have developed machine learning models based on Korean document and applied these classifiers to movie reviews to extract emotional features of various movie genres. We used Multinominal Naive Bayes model which shows best classification accuracy. Employing this model, we performed factor analysis on the movie review dataset distributed by "Naver" for 100 movies. The experimental result shows that 'vigor' and 'depression' are opposite emotions. Meanwhile, 'friendliness' positively influences on movie rating, by contrast, 'anger', 'confusion' and 'fatigue' affect negatively. Also we have found 4 movie genres which has single emotional feature as follows: horror-'anxiety', comedy-'vigor', romance/love-'friendliness' and crime-'confusion'. Finally, cross-corpus evaluation using "Twitter" and "Naver" movie review dataset demonstrates that our proposed method can be applicable in practical applications over various data attributes.

주제어