Constructing Behavioral Representation of Computational Thinking based on Event Graph: A new approach for learning analytics

Abstract

At present, the mainstream means of measuring and assessing computational thinking are test papers and scales, which enable us to measure the level of students' computational thinking as a whole, but fail to represent the inner laws and thinking processes, thus cause difficulty on providing appropriate guidance. In this study, data embedding technology was adopted to collect data on students' operational behavior in answering questions, and learning analytic techniques (i.e., semantic analysis, lagged sequence analysis, etc.) were used to evaluate computational thinking. Students with higher- and lower-level of computational thinking were compared on their behavioral patterns in a seriel events in the computational thinking training process. An event graph was constructed to represent the computational thinking behaviors based on logical event relationships. This study provides insights on a novel approach of analyzing computational thinking from the procedural perspective, which could be helpful for understanding how the thinking path is structured during problem-solving process.