摘要：Computational thinking (CT) is an approach that applies the fundamental concepts of computer science to solve problems, design systems, and understand human behavior, which can help students develop lifetime learning and generate new topics. It has been the elements of competency expected of the next generation of talents. However, the current research on computational thinking evaluation is still at a relatively weak stage. The existing related evaluation research is still limited to traditional curriculum evaluation methods. Therefore, the training effect of computational thinking cannot be well quantified, and the characteristics of students cannot be further explored. In this work, we propose a three-way decision model for improving computation thinking. We first developed a system of evaluation metrics, including five specific primary indicators and several secondary indicators. Next, the weight of each indicator was determined by applying an expert similarity measure, consequently getting the best metric sequence. We employ a grey correlation analysis to calculate the distance of each test result from this optimal sequence. Then, we trisect the set of testers based on the distance to build three regions of high score sequences, medium score sequences, and low score sequences inspired by the three-way decision. We can then exploit these rules on target students in the relatively low regions to improve their computational thinking. An example analysis illustrates the effectiveness and applicability of the method. This article provides a solid theoretical basis for improving students’ computational thinking ability. Teaching administrators can conveniently formulate computational thinking teaching strategies, and timely warning and intervention for students with poor computational thinking ability can effectively improve students’ computational thinking ability. The corresponding training measures are given to students of different ability levels to achieve differentiated and personalized training.
