US 9,813,879 B2
Mobile device executing face-to-face interaction monitoring, method of monitoring face-to-face interaction using the same, and interaction monitoring system including the same, and mobile interaction monitoring application executed on the same
Junehwa Song, Daejeon (KR); Youngki Lee, Daejeon (KR); Chulhong Min, Daejeon (KR); Chanyou Hwang, Daejeon (KR); Jaeung Lee, Daejeon (KR); Inseok Hwang, Daejeon (KR); Younghyun Ju, Daejeon (KR); Chungkuk Yoo, Daejeon (KR); Miri Moon, Daejeon (KR); and Uichin Lee, Daejeon (KR)
Assigned to KOREA ADVANCED INSTITUTE OF SCIENCE & TECHNOLOGY (KAIST), Daejeon (KR)
Filed by Korea Advanced Institute of Science and Technology, Daejeon (KR)
Filed on Jun. 12, 2014, as Appl. No. 14/302,440.
Claims priority of application No. 10-2014-0045982 (KR), filed on Apr. 17, 2014.
Prior Publication US 2015/0302846 A1, Oct. 22, 2015
Int. Cl. H04W 4/08 (2009.01); H04W 56/00 (2009.01); G10L 25/48 (2013.01); G10L 17/00 (2013.01); G10L 25/78 (2013.01)
CPC H04W 4/08 (2013.01) [G10L 25/48 (2013.01); H04W 56/001 (2013.01); G10L 17/00 (2013.01); G10L 25/78 (2013.01)] 32 Claims
OG exemplary drawing
 
1. A mobile face-to-face interaction monitoring device, comprising:
a conversation group detector for scanning one or more mobile devices in a surrounding area and setting a conversation group;
a turn detector for determining turn by using volume topography created based on sound signals detected in the mobile devices in the conversation group,
wherein the turn detector comprises:
a volume topography builder for creating the volume topography based on the sound signals in the mobile devices in the conversation group, during a training period; and
a volume topography matcher for determining current turn by matching current sound signals with the volume topography, after the training period; and
a meta-linguistic information processor for extracting meta-linguistic context of interactants in the conversation group based on the turn;
wherein the turn is a sound-signal segment where an interactant in the conversation group starts and ends the interactant's speech, and
wherein the volume topography is created by using a signature or feature vector of volume value(s) for the sound signals detected,
wherein the meta-linguistic context of interactants in the conversation group is extracted by extracting features of the turn and determining the features of the turn categorized as:
individual for each of the interactants, including speaking time, a quantity and duration of speech and non-speech turns, and associated statistics, for each of the interactants;
relational among the interactants, including orders and frequencies of exchanges of the turns and ratio thereof, among the interactants in the conversation group; and
cumulative for one or more conversation sessions and the conversation group, including a number of the interactants in the conversation sessions, a quantity and duration of the conversation sessions, and a quantity and duration of the turns in the conversation sessions and the conversation group.