[4E2-OS-19a] OS-19 (1)

[4E2-OS-19a-01] Analysis of effective features for estimating group performance using multimodal information

〇Sotaro Toya¹, Go Miura¹, Shogo Okada¹ (1. Japan Advanced Institute of Science and Technology)

Keywords:Social signal processing, Multimodal interaction

This paper focuses on developing a model for estimating the quality of discussion using multimodal features and investigating when the key multimodal features which contribute to estimate the quality is observed while the discussion. For this purpose, we use a group meeting corpus including audio signal data of participants observed in 30 meeting sessions. Also, four annotators watch conversation transcripts and annotate the score about quality of discussion using product dimension which is a sociological criteria. We extracted various kinds of features such as vocabulary,topic and dialogact. Regression models are trained to infer the annotated score from these features using linear support vector regression.