Designing for the conversational agent based on the user's mental model expressed in drawings

Abstract

Conversational agents (CAs) become more popular and useful at home. Such CAs are being developed closer to humans with the development of natural language processing and artificial intelligence technology. Accordingly, designing a CA aligned with the users’ mental model becomes a more important part of designing the voice user interface (VUI). Since the VUI is a voice-mediated interface, users naturally form a mental model of the CA’s CA through the voice. That mental model affects users’ interaction and expectations with CAs while using a VUI. However, the CA design guide for CA, which is currently used in the field of practical use, lacks a user-centered perspective. Therefore, this study first understood the perception of users through the drawing study. For this purpose, the design factors that affect users’ mental model of CA were derived through qualitative analysis of collected drawings and interviews. To understand the meaning of the design factors and the relation between them, designer workshops were conducted and further to explore how the factors derived from the user's mental model can be used and the relationship of each factor in the actual design process. As a result, this study revealed the relationship between the two design factors was drawn. Furthermore, this study found that environmental factors could be the possibility of new design considerations needed to design CA.