Construction and Research of Personalized University English Learning Platform Based on Recommendation Algorithm

Abstract With the rise of “Internet + Education”, the number of university English learning platforms has increased dramatically. Based on the original model of English teaching, this paper builds a university English learning platform. The K-Means algorithm is used to construct student profiles and cluster the student groups. Student portraits are added to the user-based collaborative filtering algorithm, and Pearson’s similarity, learning resource weights, and rating differences are integrated to propose an improved version of Pearson’s similarity to calculate student similarity, which ultimately forms an improved version of the collaborative filtering algorithm. HR and NDCG evaluation indexes are used to evaluate the performance of different models, when the number of recommendations N takes the value of 20, the improved version of collaborative filtering algorithm compared with collaborative filtering algorithm and deep neural network, in the HR evaluation indexes improved by 0.06, 0.08, in the NDCG evaluation indexes improved by 0.003, 0.01, respectively. Comparison concludes that improved collaborative filtering algorithm is more advantageous in dealing with the user sequence data is more advantageous, compared with the collaborative filtering algorithm and deep neural network performance is better, for the English learning platform recommendation algorithm module design to provide basic data information and directional guidance.