dc.description.abstract | English, as one of the globally recognized languages, has become increasingly important in the era of globalization. Achieving fluency in a second language requires time, continuous learning, and active practice. Adaptive spaced repetition methods have positively affected second language learning, enhancing learners′ long-term memory and learning quality. Through knowledge tracing algorithms, it is possible to model learners′ knowledge trajectories, understand their current grasp of knowledge, and assist them in practicing and reviewing areas of weakness.
This study expands on previous knowledge tracing algorithms and develops the Tagorithm knowledge tracing algorithm. It recruited 61 participants from two universities in northern Taiwan, with 29 in the control group and 32 in the experimental group. During practice sessions, the control group randomly selected questions from a question bank for practice and review, while the experimental group received practice and review recommendations based on Tagorithm′s predicted knowledge mastery. The experiment lasted 12 weeks, with pre-tests and post-tests conducted in the first and twelfth weeks, including English proficiency tests, online self-regulated learning questionnaires, and self-directed learning questionnaires. After the experiment, the English proficiency pre-tests and practice records were used as the training set, and the English proficiency post-tests were used as the test set to evaluate our proposed model′s predictive ability.
The research findings indicate that the Tagorithm knowledge tracing algorithm accurately predicts learners′ mastery of various knowledge points. The experimental group, which received practice recommendations based on the predictive results, showed significant improvement in learning outcomes and significantly reduced response times. Overall, compared to the control group′s random question approach, the Tagorithm knowledge tracing algorithm efficiently improved learning outcomes for learners and provided a visualized dashboard displaying the mastery levels of various knowledge points, enabling learners and administrators to understand learning progress. Following the experiment, there were no significant differences in online self-regulated learning and self-directed learning abilities between the two groups. Finally, based on the interview results, participants appreciated the algorithm′s professionalism and acknowledged the benefits of adaptive spaced repetition mechanism in helping them efficiently review and improve their English proficiency. | en_US |