Stretching out Emotion Research : From Data Collection to Modeling, Analysis, and Application
- 주제(키워드) Emotion analysis , Emotion classification , Psychological emotion model , Deep learning , Consumer needs analysis
- 발행기관 아주대학교
- 지도교수 한경식
- 발행년도 2021
- 학위수여년월 2021. 2
- 학위명 석사
- 학과 및 전공 일반대학원 인공지능학과
- 실제URI http://www.dcollection.net/handler/ajou/000000030657
- 본문언어 영어
- 저작권 아주대학교 논문은 저작권에 의해 보호받습니다.
초록/요약
Extensive research has been conducted to develop emotion classification models as a way to effectively detect and analyze emotions; however, there is still room for improvement because of (1) reliability issues of emotion labels, (2) small amount of reliable data, and (3) lack of model application. This paper presents emotion research that addresses these issues. We used a large-scale, emotion-labeled text dataset (924,827 online posts) directly specified by the authors and evaluated its validity through comparisons with other representative emotion datasets. The emotion classification model yielded performance up to 81% accuracy. We applied our model to two popular social networking sites, Reddit and Yelp, and evaluated feasibility and challenges to be considered in the application of emotion modeling. Especially, our study results highlight the ambiguity of the love emotion, and we discuss how to deal with it from theoretical perspectives. Finally, we present a case study of using emotion models to understand consumer needs.
more목차
1 Introduction 1
2 Related work 4
2.1 Psychological models of emotion 4
2.2 Emotion analysis in text 5
2.3 Consumer needs analysis on online review based on emotion analysis 7
3 Research procedure 8
4 Emotion classification model 9
4.1 Data collection and preprocessing 9
4.2 Parrott's emotion model 9
4.3 Emoji 11
4.4 Data validation 12
4.5 Modeling 17
4.5.1 Baseline Model: LSTM 18
4.5.2 Advanced Model: BERT 18
4.6 Model evaluation 19
5 Model applicability evaluation 21
5.1 Emotion labeling survey from AMT 21
5.2 Qualitative analysis of survey results 22
6 Model application case study: customer needs analysis 25
6.1 Influential user detection 25
6.2 Consumer needs analysis 27
7 Discussion 31
7.1 Study summary 31
7.2 Emotion classification model 31
7.2.1 Classification model with emoji data 31
7.2.2 Insights from model applicability evaluation 32
7.2.3 The ambiguity of Love 32
7.3 Limitation and Future work 34
8 Conclusion 35
Reference 36
