In view of the obvious differences in learning rhythms, the homogeneity of review arrangements and the instability of learning output per unit time in English vocabulary learning, this paper proposes a vocabulary learning efficiency improvement method based on K-means clustering. Based on the behavioral data collected by the online learning platform, such as login frequency, learning duration, spelling accuracy, context question completion rate, review interval and forgetting backoff times, this paper constructed learners ‘multi-dimensional feature vectors, and completed learners ‘grouping after standardized processing. On this basis, it extracted vocabulary mastery preferences and generated vocabulary learning paths for individual differences. The experimental results show that the accuracy of vocabulary test in the experimental group reaches 83.6%, which is 8.8 percentage points higher than that in the control group. The number of words mastered per unit time was increased from 18.9 /h to 24.7 /h, the 7-day retention rate was increased by 10.7 percentage points, and the average review response time was shortened to 19.8 minutes. The results show that K-means clustering can effectively identify the structural differences of learners in learning engagement, memory retention and task response, and improve the efficiency of English vocabulary learning through vocabulary task scheduling and review node optimization.
