Task-oriented Analysis on Debugging Process Based on Eye Movements and IDE Interactions

Understanding debugging behaviors is a classical topic in various research areas. Most of the existing work focused on the comparison between different groups (e.g. experts and novices). As to programming education, the details of debugging process are vital for teachers to conduct individual diagnosis. In this paper, we focus on this requirement, and perform a measurement based study on student’s debugging process. We take C programming in Eclispe IDE as the scenario, and measure the student’s eye movements and IDE interactions. A analysis method is proposed to interpret the obtained multi-modal data. The event of running program in IDE is adopted as the division indicator for debugging tasks. With this task organization model, our propose a task-oriented analysis method, which divides the data analysis work into two levels, the tasks focusing the relevant visual AoI and the actions in the same task. We further investigate the temporal evolution in multiple aspects, including the migration of task focus, the distribution of cognitive load, the time consumption of various actions, and the visual attention on relevant AoIs. Two students, who had the similar overall debugging performance, are selected as our study cases. The results show that, our method can not only recover the details of their temporal progress, but also infer their main difficulty in the debugging task.