Improving Language Models for Emotion Analysis: Insights from Cognitive Science
Constant Bonard, Gustave Cortal
TL;DR
This paper tackles fragmentation in NLP emotion analysis due to divergent annotation schemes and narrow benchmarks. It argues for a cognitively informed approach that unifies emotion theories (via an integrated framework) and applies a detective-analysis perspective from cognitive pragmatics to annotation and evaluation. Key contributions include proposing steps toward a unified annotation scheme, leveraging prompting and environmental interaction to improve knowledge use, and developing richer benchmarks that reflect the full complexity of human emotion. The work aims to align NLP emotion understanding more closely with human affective science, enabling more generalizable models and more meaningful cross-task comparisons.
Abstract
We propose leveraging cognitive science research on emotions and communication to improve language models for emotion analysis. First, we present the main emotion theories in psychology and cognitive science. Then, we introduce the main methods of emotion annotation in natural language processing and their connections to psychological theories. We also present the two main types of analyses of emotional communication in cognitive pragmatics. Finally, based on the cognitive science research presented, we propose directions for improving language models for emotion analysis. We suggest that these research efforts pave the way for constructing new annotation schemes, methods, and a possible benchmark for emotional understanding, considering different facets of human emotion and communication.