ABSTRACT Online learning has become a part of daily life. However, it has fostered a tendency to shallow reading. Notably, little attention has been paid to precisely identifying shallow reading behaviors using quantified education big data in favor of methods such as conceptual descriptions, opinions, and questionnaires, resulting in an imprecise analysis of its impact on learning and a lack of strategies to mitigate its causative factors. This study addresses the issue by introducing a behavior-identifying model grounded in measurable data to precisely identify shallow reading behaviors. Initially, three key dimensions of shallow reading behaviors were proposed, along with corresponding mathematical indicators. Subsequently, the k-means algorithm and Pearson correlation coefficient were employed to evaluate the relationship between shallow reading and learning performance, underscoring its significance. The analysis showed that shallow reading behaviors are highly associated with poor learning performance, thus hindering critical reading. These findings further validate the predominant influence of reading speed on learning performance. Consequently, a strategy was developed to intentionally decrease reading speed. Following that, an experiment was conducted to assess the effectiveness of this strategy in mitigating shallow reading. The results revealed improved learning performance and a significant reduction in shallow reading behaviors without increasing cognitive load.
Read full abstract