Abstract
The accelerated development of the metaverse has turned the virtual worlds into immersive socio-digital worlds in which the dynamics of the human interaction are closer to the real world. Nevertheless, the lack of emotional intelligence in the existing systems prevents their capacity to facilitate the meaningful and context-sensitive interaction. The given paper suggests the paradigm of Affective Metaverse Computing, which incorporates the multimodal emotion intelligence into human-centered virtual interaction systems. Using the recent technologies of affective computing, deep learning, and multimodal data fusion, the framework proposed will allow recognizing, interpreting, and responding to the emotions of the user in real-time using visual, auditory, textual, and physiological data.