Voice User Interfaces (VUIs) pose challenges for users to learn the system’s supported features and commands. Users often rely on trial-and-error to navigate VUI dialogues and complete desired tasks. The order in which users try different commands contains vital information about how they learn. In this paper, we explore using sequence analysis techniques to reveal the patterns of tactics our participants (n = 50) used when interacting with an unfamiliar multi-modal VUI. We present the results of our sequence analysis, the top sequences used, and a cluster analysis of our participants on their usage of the top sequences. Our results indicate participants initially struggled with understanding the acceptable utterance structure and entities more so than utterance keywords. Additionally, we discuss our participants’ behavior differences and discuss the usefulness of this methodology of sequence analysis for HCI VUI research.
|Title of host publication
|CUI 2021-3rd Conference on Conversational User Interfaces
|Number of pages
|Published - 2021