PhD Seminar • Human-Computer Interaction — A Four-Component Conceptual Model of the Conversational Agent Work Process

Anastasia Kuzminykh, PhD candidate
David R. Cheriton School of Computer Science

As the popularity of conversational agent systems grows, there is a pressing need to understand end-users’ perceptions of the types of tasks and the interactions that these agents can and should support. 

To address this problem, we develop an empirically grounded conceptual model of the conversational agent work process, which includes four components: 

1. a classification of interaction types,
2. a corresponding typology of conversational agent tasks,
3. two task execution flows associated with suggested interaction types, and
4. a classification of users’ concerns mapped onto the task execution flows. 

The conceptual model both broadens our understanding of the possible interactions, tasks, and user concerns, and provides guidance for future refinements to this ever-more-pervasive technology.