Tailoring Conversational UX through the Lens of Dialogue Complexity

Abstract

Interactions with conversational systems take diverse forms across applications and individual users. In order to develop intelligent systems that can accommodate these diverse conversational user experience (UX) needs and preferences, meanwhile balancing implementation costs, we propose a computation-driven approach to profile conversational interactions by measuring dialogue complexity of user inputs in multiple dimensions. To inform system adaption designs, we propose to conduct comparative conversation analysis and user experiments, in order to develop classified UX guidelines for conversational interactions with different complexity profiles.