Accurate recommendation and adaptation model of college English teaching and learning resources assisted by machine learning

Under the background of digitalized college English teaching, this study constructed a machine-learning-assisted teaching resource recommendation and adaptation model based on learner features, resource features, and learning feedback. A quasi-experimental design was conducted with 124 first-year non-English majors in a compulsory College English II course, including 63 students in the experimental group and 61 students in the control group. The dataset contained 186 teaching and learning resources and 4,512 valid learner-resource interaction records. The proposed model achieved Precision@5 = 0.73, Recall@5 = 0.66, NDCG@5 = 0.78, F1@5 = 0.69, RMSE = 0.64, and MAE = 0.51, outperforming common baseline methods. After the eight-week intervention, the experimental group’s total score increased by 5.6 points, while the control group increased by 1.7 points; the between-group gain difference was significant, 95% CI [2.84, 4.96], p < 0.001, Cohen’s d = 1.31. The experimental group also showed higher task efficiency, shorter average task time, more continuous learning days, and a higher active resource click ratio. These results indicate that the model can support more accurate resource recommendation, improve short-term English learning outcomes, and enhance learning efficiency and engagement under the present teaching conditions.