In fields such as healthcare, education, counseling, customer service, and interactive storytelling, persona-based dialogue agents have made significant advancements in emotion-sensitive role simulation. However, there are two major gaps in current research: first, persona-based dialogue systems often encode emotions as static traits or surface-level stylistic cues; second, affective dialogue research primarily focuses on empathetic response generation towards users rather than modeling the agent persona's own evolving emotional state.
Thus, trigger-driven emotional evolution within a character remains underexplored.
To address this gap, we draw inspiration from the Component Process Model (CPM), a psychological theory that views emotion as a dynamic process shaped by the appraisal of external events. We propose CPM-MultiAgent, a CPM-grounded emotion evolution multi-agent framework to support emotional changes in persona-based dialogue. Unlike treating a character's emotion as a fixed attribute, CPM-MultiAgent represents it as a latent state continuously reshaped by dialogue triggers.
Through affective trigger extraction, CPM-based collaborative appraisal, and emotion state updating, the framework enables more emotionally consistent role simulation in multi-turn interactions. Experiments including baseline comparisons, ablation studies, human evaluations, and case analyses demonstrate that CPM-MultiAgent effectively models dynamic emotional evolution in emotionally sensitive role-simulation settings.
Blogger's Review: The CPM-MultiAgent framework proposed in this paper opens a new perspective in the field of affective computing by dynamically modeling the emotional evolution of characters, enhancing the emotional consistency of dialogue systems. This approach holds great potential for applications in scenarios requiring long-term emotional simulation and deserves further exploration and optimization.