Interactive Mouse Stream as Real-Time Indicator of User's Cognitive Load.

User interaction and multimodal behavior have been argued as viable indicators of cognitive load. We extend this idea by exploring interactive mouse data stream and implementing sliding windows technique to detect cognitive load variation in real-time. This work-in-progress reports successful load change detections resulting from applying our unique algorithm to data streams of mouse interactivity features from twenty seven subjects. Unique contribution here includes learning from mouse interactive stream and a sliding window technique for cognitive load detection in real-time. This technique is currently being enhanced to process learning from multimodal user interaction streams.