Citation Context

...oadcast Data, Meeting Recordings. I. INTRODUCTION The computing community is making significant efforts towards the development of automatic approaches for the analysis of social interactions (see [1]=-=[2]-=-[3] for extensive surveys of the domain). This is not surprising as social interactions are not only one of the most important aspects of our everyday lives, but also an ubiquitous subject in multimed...

Citation Context

...efforts have been done, to the best of our knowledge, towards the automatic inference of socially and emotionally relevant information from the way people use, organize and share their physical space =-=[66, 67]-=-. In contrast, this subject has been extensively investigated in human sciences where the spatial arrangement of people in social encounters has been shown to be a reliable evidence of the social phen...

Citation Context

...categorized into a growing research domain called social signal processing [38]. Complete introduction of this domain is beyond the scope of this paper, but a recent literature survey can be found in =-=[39]-=-. We proposed an SNA-based approach to analyze movies in [20]. Leading roles and corresponding communities can be automatically identified by checking the social relationships between characters. Howe...

Citation Context

...tep further, we might use inductive (or other) methods to infer specifications like the one for β from the behaviours we observe in message passing (this is similar to ideas such as social signalling =-=[7]-=-) or, following the ideas in Harel [4], we might be able to build devices with the ability to infer some forms of role definition from user requirements expressed via the operation of the device itsel...

Citation Context

...nverbal Behavior 1. INTRODUCTION Recently, there has been a growing interest in the analysis of nonverbal social signals using machine learning and signal processing by the computer science community =-=[1, 2, 3]-=-. The understanding of behavior in teams, that can be either cooperative or competitive, is of vital importance for organizational management [4]. Understanding individual behavior and estimating when...

Citation Context

...d revealing about socially relevant information. Low level speech features like spectral cues and prosody can be used to infer different signals from speech, such as motivation, empathy and dominance =-=[22]-=-. In [23], prosody is used to predict personality traits attributed to speakers. The authors used the well-known Big Five personality model to assess extraversion, agreeableness, conscientiousness, ne...

Citation Context

... intelligence [12] [13]. Due to it’s necessity, social signal processing has became a new domain that aims to automatically sense and understand human social interactions through machine analysis [4] =-=[14]-=-. One of the earliest works in this domain focused on social signal detection for predicting the outcome of dyadic interactions such salary negotiations, hiring interviews, and speed-dating conversati...

Citation Context

...is kept low, as the only specialized hardware required are the networked bluetooth scanners (∼ $50−$100 each). Navigation and location sharing has been explored extensively in the past, both in urban =-=[1, 2, 3, 9, 11, 12]-=- and in a potentially indoor environment [4, 5, 6, 7, 8, 10], but in all cases user location information needs to be stored in a centralized database before it can be shared among users. This limitati...

Citation Context

...t on are those emerging in “face-to-face interaction”. With specific reference to non-verbal cues, examples are facial expressions, eye gaze, head gestures and orientation, posture, body expressivity =-=[17]-=-. An examplar work is the study by Kapoor et al. [4], who proposed a method to detect frustation in users interacting with a learning companion agent based on analysis of multimodal non-verbal behavio...