Conversation Envisioning Framework for Situated Conversation

摘要

This paper introduces virtual reality conversation envisioning (VRCE) framework as an effective approach for analyzing situated conversations. The goal of VRCE is to raise understanding of shared information, facilitate common ground formation and promote smooth communication, especially in cross-cultural interactions. In this method, a situated conversational scenario is reconstructed in a VR environment to enable detailed analysis from first and third person view, empowered by flexible traverse in the time dimension. This framework allows participants and meta-participants (observers) to actively engage in the envisioning process in VR. A conversation description language (CDL) is introduced for encoding the obtained interpretations and developing a conversation envisioner. We focused on a bargaining scenario as a situated conversation with rich cultural practices. Preliminary experiments with this scenario indicated the effectiveness of VRCE to achieve better reasoning about the situation and received positive participant feedback.