Processing and Understanding Moodle Log Data and Their Temporal Dimension

Abstract

The increased adoption of online learning environments has resulted in the availability of vast amounts of educationallog data, which raises questions that could be answered by a thorough and accurate examination of students’ onlinelearning behaviours. Event logs describe something that occurred on a platform and provide multiple dimensionsthat help to characterize what actions students take, when, and where (in which course and in which part of thecourse). Temporal analysis has been shown to be relevant in learning analytics (LA) research, and capturingtime-on-task as a proxy to model learning behaviour, predict performance, and prevent drop-out has been thesubject of several studies. In Moodle, one of the most used learning management systems, while most events arelogged at their beginning, other events are recorded at their end. The duration of an event is usually calculated asthe difference between two consecutive records assuming that a log records the action’s starting time. Therefore,when an event is logged at its end, the difference between the starting and the ending event identifies their sum,not the duration of the first. Moreover, in the pursuit of a better user experience, increasingly more online learningplatforms’ functions are shifted to the client, with the unintended effect of reducing significant logs and conceivablymisinterpreting student behaviour. The purpose of this study is to present Moodle’s logging system to illustratewhere the temporal dimension of Moodle log data can be difficult to interpret and how this knowledge can be usedto improve data processing. Starting from the correct extraction of Moodle logs, we focus on factors to considerwhen preparing data for temporal dimensional analysis. Considering the significance of the correct interpretation oflog data to the LA community, we intend to initiate a discussion on this domain understanding to prevent the loss ofdata-related knowledge.