Design of Intelligent Recommendation System for English Learning Path of Textile Materials Based on Graph Neural Network

With the rapid development of intelligent information transmission and semantic communication technologies, personalized knowledge delivery and adaptive learning have become increasingly important in engineering-oriented educational systems. However, most English learning platforms employ unified learning paths that fail to capture learners’ individual characteristics, professional backgrounds, and knowledge dependencies, particularly in the specialized domain of textile material terminology. To address this issue, this paper proposes an intelligent recommendation system for textile material English learning paths based on Graph Neural Networks (GNNs). By incorporating the ASTM D123-21 textile terminology standard, a student–knowledge heterogeneous graph is constructed to jointly model learner attributes, professional knowledge, and multidimensional relationships. A Multilevel Graph Attention Aggregation (MGAA) mechanism is further developed to integrate domain semantics with dynamic learning behaviors, while a Heterogeneous Graph Convolutional Network (HGCN) combined with reinforcement learning enables adaptive path optimization according to learner feedback. Experimental results demonstrate that the proposed framework improves path recommendation accuracy, knowledge coherence, and professional adaptability to 96.81%, 0.88, and 0.93, respectively, and achieves a 14.4% improvement in learning efficiency compared with the Knowledge-aware Graph Attention Network (KGAT). The proposed method provides an effective solution for personalized professional education and offers valuable insights into knowledge-driven semantic information transmission, intelligent recommendation, and adaptive communication architectures for future engineering applications.