Detection of division of labor in multiparty collaboration

Abstract

In the research field of human-computer interaction, there are many approaches to predicting interactive roles, e.g., conversational dominance or active participation. Although interactive roles have been predicted for entire tasks, little attention has been given to evaluating how such roles are reorganized during a task. This paper explains how to construct a model for predicting emergent division of labor and the reorganization of labor in multiparty collaboration using verbal and nonverbal cues. To build the model, we adopted stepwise multiple-regression analysis, which is a type of statistical model analysis, using both behavioral data and third-party evaluations. We confirmed useful verbal and non-verbal parameters for predicting interactive roles and their reorganization through this model. © 2013 Springer-Verlag Berlin Heidelberg.