Emotion Overflow: an Interactive System to Represent Emotion with Fluid

This paper presents Emotion Overflow, a novel system that transforms textual emotional input into interactive fluid visualizations. Utilizing a GPT model, our approach extracts emotion attributes from text and maps them onto fluid characteristics, creating a dynamic visual representation of emotional states. The system focuses on valence and arousal as key emotion attributes, translating them into fluid properties such as color, variance over time, and responsiveness. We demonstrate the application through two interactive experiences: a direct emotion visualization and an emotion guessing game. This work contributes to the field of emotional expression in human-computer interaction by offering a fluid, intuitive interface for exploring and communicating complex emotional states.