Earticle

초록

영어

Emotional digital humans have powerful applications in various fields due to their ability to respond to users' emotional needs. The realization of emotional digital humans requires a complex technological approach that tracks emotions and generates responses that match the flow of conversation. We aim to enable digital humans to respond more precisely and effectively emotionally, so that they can be applied to various services through emotional and intuitive conversations with people. To this end, we developed a model that analyzes emotional changes occurring during conversations by combining the latest deep learning-based emotion analysis model and natural language processing techniques. The model proposed in this study consists of a emotion analysis module for multimodal data, a transformer-based context recognition module for contextual emotion recognition in conversations, and an adaptive emotional response module based on reinforcement learning that reflects user feedback. As a result of our research, we confirmed that a multimodal model that integrates text, voice, and facial expression data has higher accuracy and F1 score in emotion recognition than a single modal model. We confirmed emotional changes according to the conversation stage, which shows that digital humans should adjust their emotional responses according to the progress of the conversation. We expect that this study will provide insights into building an emotion analysis framework essential for digital humans with emotional intelligence.

목차

Abstract
1. Introduction
2. Related Works
2.1 Digital Human and Emotion Analysis
2.2 Conversation-based Emotion Analysis Techniques
3. Research Methods
3.1 Data Collection and Preprocessing
3.2 Feature Extraction and Preprocessing of Emotional Expression
3.3 Model Design
4. Experiments and Results
4.1 Experimental Environment and Conditions
4.2 Model Performance Comparison and Analysis
4.3 Analysis of Conversation Features according to Emotional Changes
4.4 Summary of Experimental Results and Findings
5. Discussion
6. Conclusion
Acknowledgement
References

키워드

저자

• Jee Young Lee [ Associate Professor, Department of Software, SeoKyeong University, Korea ] Corresponding Author

참고문헌

발행기관

간행물

표지보기 관심저널 등록

• 간행물명

  International Journal of Advanced Culture Technology(IJACT)

• 간기

  계간

• pISSN

  2288-7202

• eISSN

  2288-7318

• 수록기간

  2013~2025

• 등재여부

  KCI 등재

• 십진분류

  KDC 600 DDC 700