Methodology of Affective Intervention Design for Intelligent Systems

摘要

This paper concerns how intelligent systems should be designed to make adequate, valuable and natural affective interventions. The article proposes a process for choosing an affective intervention model for an intelligent system. The process consists of 10 activities that allow for step-by-step design of an affective feedback loop and takes into account the following factors: expected and desired emotional states, characteristics of the system and available input channels for emotion recognition. The proposed method is supported with three examples of intelligent systems that have used the process to design or to describe affective intervention. Two of them are case studies of affective video games that were designed with the proposed approach and evaluated against the non-affective versions. Another example described in the article is the affective tutoring system Gerda, which exemplifies a more complex and sophisticated intervention model.