Towards Collaborative Conversational User Interfaces for Team Training

Abstract

Conversational interfaces are quickly becoming a prominent type of intelligent user interface. However, much of the current work has focused on dyadic interaction between a single user and a single conversational agent. The aim of my PhD thesis is to examine how conversational interfaces can be adapted to support humans engaged in a team task. My work so far has identified the key types of functionality such conversational interfaces should entail and via a series of a Wizard of Oz studies I have validated these findings, and obtained the data needed to train a conversational model to support human-artificial agent teaming. My aim at the Doctoral Consortium would be seek guidance from and form research collaborations with my peers on a general approach to designing, training and evaluating collaborative conversational agents for human-machine teaming.