Mining Effective Learning Behaviors in a Web-Based Inquiry Science Environment

Analyzing learners’ learning behaviors helps teachers understand how learning behaviors of learners influence learning performance. To determine which learning behaviors influence learners’ science-based inquiry learning performance, this work develops an xAPI (Experience Application Programming Interface)-based learning record store module embedded in a Collaborative Web-based Inquiry Science Environment (CWISE) to record detailed data about students’ learning processes. This work discusses whether the significant correlation and cause-effect relationship among science inquiry competence, learning time, and learning performance exist, and examines whether remarkable shifts and differences in the learning behaviors of learners with different learning performances and inquiry competences exist by using sequential pattern mining and lag sequential analysis. The results demonstrate that inquire ability, total learning time in the designed inquiry learning course, and learning time in an inquiry buoyancy simulation experiment are positively correlated with learning performance and can predict learning performance, and the learning time in the inquiry buoyancy simulation experiment appears to be the most significant predictor. The results of lag sequential analyses indicate that learners with high learning performance and high inquiry competence re-adjust hypotheses after performing an inquiry buoyancy simulation experiment, while learners with low learning performance and low inquiry competence lack this critical inquiry learning behavior. This study presents a systematic analysis method to insight the effective learning behaviors in a web-based inquiry learning environment based on mining students’ learning processes, thus providing potential benefits in guiding learners to adjust their learning behaviors and strategies.