Performance Evaluation Through Log Analysis for Interactive Simulations

This poster presents ideas for implementing computer-based performance evaluation through log data analysis obtained from students' interactions with online simulations. We developed an interactive learning material that simulates a schematic experiment of the photoelectric effect using GeoGebra, a dynamic mathematics software, and STACK, an online assessment system, on Moodle LMS. Students can measure the photocurrent by adjusting the anode voltage and selecting appropriate wavelengths of the emitting light as well as the target cathode metals. These interactions produce time-series log data, which can be compared with an ideal experimental procedure and analyzed to evaluate the appropriateness of students' performance. As our simplest option for the comparison mechanism, we use regular expressions to define the ideal operational sequences and conduct pattern matching to detect if such sequences appear in the students' log data. More elaborate methods to better handle the time-series property of the log are also being developed. By focusing on not only the final states but also the process of interactions, the implementation provides rich information on the quality of learning, which is hard to obtain in ordinary assessments.