Context-based natural language participant modeling for videoconference focus classification

摘要

A video conferencing system includes a processing device and a memory device and is configured to manage a video conference to manage a video conference feed to a plurality of participant devices. The system includes an interface control module that receives a plurality of video feeds and audio feeds associated with participants to the video conference, and generates a video conference feed including a video conference interface for being displayed to the plurality of participants. The system also includes a context analysis module that analyzes the audio feeds to identify contextual information associated with the conversation. The system further includes a participant modeling module that calculates a topic relevance score for each of the plurality of participants based on the contextual information. During a video conference, the interface control module emphasizes a video feed of a participant associated with a topic relevance score greater than a threshold.