Research on Intelligent Recommendation System of English Teaching Resources Driven by Big Data

This paper proposes an intelligent recommendation system of English teaching resources driven by big data, which combines the advantages of collaborative filtering and deep learning (especially neural networks). The system first uses collaborative filtering algorithm to mine similarities between students from students' historical behavior data, predicts the resources that students may be interested in, and then further optimizes the recommendation through deep learning model to learn students' potential interest changes and complex learning patterns. Experimental results show that the recommendation accuracy of the recommendation system on multiple standard data sets (such as LibriSpeech, TIMIT, etc.) reaches 91%, which is 15% higher than that of traditional collaborative filtering methods. In addition, the system also performs well in real-time performance, and the recommendation response time is reduced by about 20% compared with traditional methods, which can meet the real-time recommendation needs in large-scale online education environments. Through experimental comparative analysis, the model proposed in this paper shows obvious advantages in recommendation accuracy, real-time and system adaptability, especially in complex learning environments (such as background noise, differences in students' learning habits, etc.). It is also more robust.