Utilizing Human-Computer Interaction Data to Extract User Interests from Web-based Learning Systems

We propose a novel approach by utilizing human-computer interaction data to extract user interests from web-based learning systems. The proposed approach is based on user access and a wearable context recognition system. Under our approach, a series of screen images are captured by an imaging device worn by users engaged in web-based learning. These images help in detecting specific vendor logo information, which is then used to deduce the web-based learning context. The compiled history of recognized context and learning access is finally compared to extract user interests. Experimental results show that the proposed approach is robust, and can identify the relevant context in 96% of cases. The proposed method was successfully applied to 16 users for 1 h of learning time to extract high and low interest topics from web-based learning systems.