In recent years, learning analytics and intelligent recommendation technologies have continuously entered the foreign language education scene, making English autonomous learning gradually shift from experience support to data driven. Aiming at the problems of coarse granularity of resource push, insufficient recognition of learning status, and disassociation between recommendation results and process feedback in existing English learning platforms, this paper proposes an English self-regulation learning recommendation model based on learning behavior data. This paper extracts multi-dimensional features from login, browsing, practice, test, error correction and review behaviors, constructs learning state representation, resource matching and dynamic ranking mechanism, and further designs a comprehensive management and personalized recommendation system for English learning resources. Experimental results show that the proposed model achieves 0.412, 0.386, 0.447 and 0.521 on Precision@5, Recall@5, NDCG@5 and MRR indicators, respectively. The system still maintains good response performance under the condition of resource scale expansion and concurrent access. The results show that learning behavior data can provide effective support for English autonomous learning recommendation.
