Abstract: Being an effective communicator plays a major role in employment interviews. In this paper, we provide a computational framework to automatically predict the communication skill of a person in an interface-based interview setting. The advantage of interface-based interview setting compared to that of a face-to-face setting is, the participants get assessed without any human intervention. To this end, we have collected audio-visual recordings of 106 participants. Our auto-mated analysis includes extraction of both audio and visual behavioral cues such as prosodic, speaking activity-based and facial expression cues. Ground truth labels are derived by taking the mode of three independent judges. Our framework automatically predicts the communication skill rating using regression and classification models. We also, derive the importance of manually annotated attributes like speaking flu-ency and rate of speaking to predict the communication skill rating.
Keywords:
- Communication skills
- Interface based interview
- Social computing
