Investigating mind markers in design meetings

In the context of the European AMI1 project, “Augmented Multiparty Interaction��? more than one hundred hours of video and audio data have been collected of groups of four people engaged in a meeting with the task to design a new remote control. The collection of meetings is being annotated on several layers, ranging from speech transcripts to emotion annotation. The main purpose of this data collection within the AMI project is to train, through machine learning techniques, automatic recognizers and interpreters of the data that would be able to deliver automatic meta-data which can be used in browsing and retrieval of new recordings. Besides this technological goal, the data collection will also be useful for the social sciences, particularly for the study of small group interactions.