Emotion Analysis in NLP: Trends, Gaps and Roadmap for Future Directions

• DOI: 10.48550/arxiv.2403.01222
• Publication Date: 2024-03-02
• Authors: Flor Miriam Plaza...; Alba Curry; Amanda Cercas Curry; Dirk Hovy
• Keywords: FOS: Computer and information sciences; natural language processing, emotions; Computer Science - Computation and Language; Computation and Language (cs.CL)

Emotions are a central aspect of communication. Consequently, emotion analysis (EA) is rapidly growing field in natural language processing (NLP). However, there no consensus on scope, direction, or methods. In this paper, we conduct thorough review 154 relevant NLP publications from the last decade. Based review, address four different questions: (1) How EA tasks defined NLP? (2) What most prominent frameworks and which emotions modeled? (3) Is subjectivity considered terms demographics cultural factors? (4) primary applications for EA? We take stock trends tasks, used, existing datasets, methods, applications. then discuss lacunae: absence demographic aspects does not account variation how perceived, but instead assumes they universally experienced same manner; poor fit categories two main theories to task; lack standardized terminology hinders gap identification, comparison, future goals; interdisciplinary research isolates insights other fields. Our work will enable more focused into holistic approach modeling NLP.