What are the challenges of traditional text-based emotion recognition in affective computing?

Traditional text-based emotion recognition in OMPuting" class="internal-link">affective computing faces challenges such as oversimplification of emotions, neglect of interpersonal dynamics, and the inability to capture nuanced emotional expressions. This approach often treats emotions as fixed attributes of the text, failing to account for the variability in emotional responses influenced by individual personalities and contextual factors. Additionally, it typically focuses on the writer's sentiment without considering how the reader interprets or reacts to that sentiment, which can lead to misinterpretations of emotional intent. For instance, research has shown that individuals can exhibit ambivalence and hesitancy in their emotional expressions, where their verbal communication may conflict with non-verbal cues like facial expressions or tone of voice. A study by Kessler et al. (2020) highlighted that relying solely on text can lead to a loss of critical contextual information, which is essential for accurately understanding emotions in dyadic interactions. This underscores the need for more comprehensive models that incorporate both verbal and non-verbal signals to improve emotion recognition in affective computing.