감정은 서사 생성과 이해 모두에서 중요한 역할을 한다. 본 논문은 플루칙의 감정 모델을 기반으로 영화 대본에서 8가지 감정 표현을 분석하였다. 먼저 각 장면별 수동으로 감정을 태깅하였고, 이 때 8가지 감정 중 분노, 공포, 그리고 놀람이 가장 우세하게 나타났는데, 이는 스릴러 영화 장르를 고려할 때 의미있다고 할 수 있다. 또한, 스토리에서 긴장이 가장 고조되는 클라이맥스에서 다양한 감정이 복합적으로 나타난다고 가정하였고, 대본 상에서 3 부분의 클라이맥스를 확인할 수 있었다. 그 다음으로 파이썬 (Python) 프로그래밍 언어 기반 자연어처리 도구인 NLTK (Natural Language ToolKit)의 감성 분석 도구를 이용하여 수동 감정 태깅과 비교한 결과, 분노와 공포 감정에서 높은 일치율을, 그리고 놀람, 기대, 혐오 감정에서는 낮은 일치율을 보임을 확인하였다.
감정은 서사 생성과 이해 모두에서 중요한 역할을 한다. 본 논문은 플루칙의 감정 모델을 기반으로 영화 대본에서 8가지 감정 표현을 분석하였다. 먼저 각 장면별 수동으로 감정을 태깅하였고, 이 때 8가지 감정 중 분노, 공포, 그리고 놀람이 가장 우세하게 나타났는데, 이는 스릴러 영화 장르를 고려할 때 의미있다고 할 수 있다. 또한, 스토리에서 긴장이 가장 고조되는 클라이맥스에서 다양한 감정이 복합적으로 나타난다고 가정하였고, 대본 상에서 3 부분의 클라이맥스를 확인할 수 있었다. 그 다음으로 파이썬 (Python) 프로그래밍 언어 기반 자연어처리 도구인 NLTK (Natural Language ToolKit)의 감성 분석 도구를 이용하여 수동 감정 태깅과 비교한 결과, 분노와 공포 감정에서 높은 일치율을, 그리고 놀람, 기대, 혐오 감정에서는 낮은 일치율을 보임을 확인하였다.
Emotion plays a key role in both generating and understanding narrative. In this article we analyzed the emotions represented in a movie script based on 8 emotion types from the wheel of emotions by Plutchik. First we conducted manual emotion tagging scene by scene. The most dominant emotions by man...
Emotion plays a key role in both generating and understanding narrative. In this article we analyzed the emotions represented in a movie script based on 8 emotion types from the wheel of emotions by Plutchik. First we conducted manual emotion tagging scene by scene. The most dominant emotions by manual tagging were anger, fear, and surprise. It makes sense when the film script we analyzed is a thriller-genre. We assumed that the emotions around the climax of the story would be heightened as the tension grew up. From manual tagging we could identify three such duration when the tension is high. Next we analyzed the emotions in the same script using Python-based NLTK VADERSentiment tool. The result showed that the emotions of anger and fear were most matched. The emotion of surprise, anticipation, and disgust, however, scored lower matching.
Emotion plays a key role in both generating and understanding narrative. In this article we analyzed the emotions represented in a movie script based on 8 emotion types from the wheel of emotions by Plutchik. First we conducted manual emotion tagging scene by scene. The most dominant emotions by manual tagging were anger, fear, and surprise. It makes sense when the film script we analyzed is a thriller-genre. We assumed that the emotions around the climax of the story would be heightened as the tension grew up. From manual tagging we could identify three such duration when the tension is high. Next we analyzed the emotions in the same script using Python-based NLTK VADERSentiment tool. The result showed that the emotions of anger and fear were most matched. The emotion of surprise, anticipation, and disgust, however, scored lower matching.
* AI 자동 식별 결과로 적합하지 않은 문장이 있을 수 있으니, 이용에 유의하시기 바랍니다.
제안 방법
In order to compare the results, we first manually tagged the script of the film A Hard Day (Kkeut-kka-ji-gan-da) based on the 8 fundamental emotion types introduced by Plutchick – anticipation, trust, joy, sadness, anger, fear, disgust, and surprise [14].
Typically sentiment refers to either positive or negative emotional state, mood, or attitude, while emotion has more diverse definitions ranging from discrete emotions to continuos emotional state via distinct dimensions. In this article we focus more on emotions rather than sentiments, and propose a systematic approach to identify and analyze emotions in text.
In order to compare the results, we first manually tagged the script of the film A Hard Day (Kkeut-kka-ji-gan-da) based on the 8 fundamental emotion types introduced by Plutchick – anticipation, trust, joy, sadness, anger, fear, disgust, and surprise [14]. Two undergraduate students read the script, and then tagged emotions that could describe each scene appropriately, with the intensity range from 1 to 3.
이론/모형
We also analyzed the sentiment of the translated scenario text data using VADERSentiment tool in NLTK. We first calculated compound sentiment value of each sentence, and then select representative emotion that is closet to the compound value of each emotion in Plutchik’s wheel of emotions.
Next, we conducted a sentence-based emotion analysis using Google Translate and NLTK (Natural Language Toolkit). While there exists Korean language corpus for sentiment analysis such as KOSAC (Korean Sentiment Analysis Corpus) [15], in this paper we chose to use Google Translate in order to analyze emotions sentence by sentence in each scene. Google Translate is a web-based free translating service that is most used worldwide, and its accuracy rate is reported up to 80% from non-English to English translation [16].
성능/효과
Thus we compared the differences between the average value of human tagging emotions and the accumulated emotion values from NLTK VADERSentiment. Overall the complete matching rate (that is, zero difference) was 42.6% (average of positive emotions:38.2%; average of negative emotions: 47.0%), and the differences less than 2 (that is, differences from zero up to 1) was 68.3% (average of positive emotions: 61.0%; average of negative emotions: 75.6%). Table 2 shows the matching rate between human tagging and NLTK VADERSentiment for each emotion.
참고문헌 (17)
W. Kintsch, "Learning from text, levels of comprehension, or: Why anyone would read a story anyway?," Poetics, Vol. 9, No.1-3, pp. 87-98, 1980.
K. Oatley, "A taxonomy of the emotions of literary response and a theory of identification in fictional narrative," Poetics, Vol.23, pp. 53 - 74, 1994.
S. Keen, "A theory of narrative empathy," Narrative, Vol. 14, No. 3, pp. 207-236, 2006.
A. J. Reagan, L. Mitchell, D. Kiley, C. M. Danforth, and P. S. Dodds. "The Emotional Arcs of Stories Are Dominated by Six Basic Shapes". EPJ Data Science, 5:31, 2016.
R. Feldman, "Techniques and Applications for Sentiment Analysis," Communications of the ACM, Vol. 56, No.4, pp.82-89. 2013.
J.P. Zagal, N. Tomuro, N. and A. Shepitsen. "Natural Language Processing in Game Studies Research: An Overview", Simulation & Gaming, vol 43, no.3, pp. 356-373, 2012.
P. Ekman and W. V. Friesen. "Universals and Cultural Differences in the Judgments of Facial Expressions of Emotion", Journal of Personality and Social Psychology, Vol53, No.4, pp. 712-717, 1987.
A. Ortony, G. Clore, and A. Collins, The cognitive structure of emotions, Cambridge University Press, 1988.
D. Britz, "Implementing a cnn for text classification in tensorflow." (2015). Available: http://www.wildml.com/2015/12/implementing-a-cnn-for-text-classification-in-tensorflow/.
K-M. Ahn, Y-S.. Kim, Y-H. Kim, and Y-H. Seo, "Sentiment Classification of Movie Reviews Using Levenshtein Distance," Journal of Digital Contents Society, Vo. 14, No.4, pp. 581-587. 2013.
R. Plutchik, Emotion: A psychoevolutionary synthesis. Harpercollins College Division, 1980.
H. Shin, M. Kim, Y. Jo, H. Jang, and A. Cattle. KOSAC(Korean Sentiment Analysis Corpus), Information and Computation, pages 181-190, 2013.
E. Balk, M. Chung, N. Hadar, K. Patel, W. W. Yu, T.A. Trikalinos, and L.K.W. Chang, "Accuracy of data extraction of non-English language trials with Google Translate," Methods Research Reports, 2012.
※ AI-Helper는 부적절한 답변을 할 수 있습니다.