Human Desire Inference Process Based on Affective Computing

In order for the intelligent assistant systems to provide users with timely and appropriate assistances, most of them focus on what is considered to be the "rational" aspect of the user behaviors. However, since human desire is the fundamental driving force of human behaviors, without including users\´ desires, the system cannot provide most appropriate responses. We propose a hierarchical desire inference process based on the Bayesian Belief Networks (BBNs), that considers the affective states, behavior contexts and environmental contexts of a user at given points in time to infer the user\´s desire. The inferred desire of the highest probability from the BBNs is then used in the follow-up decision making.