Investigation on the Fusion of Multi-modal and Multi-person Features in RNNs for Detecting the Functional Roles of Group Discussion Participants

摘要

More and more companies are putting emphasis on communication skill in the recruitment of their employees and adopt group discussion as part of their recruitment interview. In our ongoing project, we aim to develop a training system that can provide advices to its users in improving the perception of their communication skill during group discussion. In order to realize this goal, a conceptual unit of communi-cational behaviors and a template of communication style are required. We propose the use of functional roles of the participants in group discussions as this unit. In order to incorporate the use of functional roles for improving the perception of participants' communication skill, the first task is automatic detection of the participants' functional roles in real-time. We previously proposed a SVM based model for this task but the results were only moderate. We expect including temporal characteristics, frame-wise interaction of modalities, and inter-person interaction can improve the classification accuracy and explored the use of RNN based networks to see the effectiveness of these factors.