Traditional English writing tutoring systems have significant deficiencies in personalized feedback, writing style diversity, and dynamic adaptation to students’ needs. This paper combines generative adversarial networks, language models, and collaborative filtering algorithms to design an intelligent tutoring system to comprehensively improve students’ English writing skills. Firstly, the generator and discriminator models in GAN (Generative Adversarial Network) are used to generate high-quality writing examples and improvement suggestions to ensure the grammatical correctness and fluency of the writing content. Secondly, the naturalness and creativity of the generated text are further improved by combining the Transformer-based language model (GPT-3), helping students understand different writing styles and techniques. To enhance personalization and dynamic adaptability, this paper uses a collaborative filtering algorithm to analyze students’ writing history and progress, adjust feedback strategies in real-time, and provide personalized writing guidance that meets students’ actual needs. Through evaluation, the average grammatical error rate of students improved by the intelligent tutoring system does not exceed 10%, and the vocabulary diversity increases by up to 20 types. This paper verifies the effectiveness of the intelligent tutoring system in improving students’ English writing skills and demonstrates the advantages of the system in personalized feedback and creative expression.
Li et al. (Wed,) studied this question.