1 Introduction

Introduces the importance of personalized reading materials in international Chinese education, points out the difficulties faced by teachers in practical teaching, elaborates on the new pathways provided by large language models to address this issue, and outlines the research content and structure of this study.

2 Research Status of Reading Material Generation in International Chinese Education

Reviews the current research status of reading material generation in international Chinese education, noting that domestic scholars have conducted studies on using generative artificial intelligence to produce texts for international Chinese teaching. For example, Xu Juan and Ma Ruiling employed ChatGPT to generate reading materials and reading questions, Zhu Yijin and Rao Gaoqi constructed example sentence libraries, Wu Qiong and Tang Gaigai guided large models to generate text resources that meet standards, and Ou Zhigang et al. emphasized generating multimodal teaching resources. Foreign research focuses on reading skill acquisition theories and the development of specialized reading materials for specific teaching contexts. Although there are perspectives on personalized learning material generation, actual tools for generating reading materials have not yet been developed. Overall, using generative artificial intelligence for reading material generation research is theoretically feasible and can meet teaching and learning needs. However, there is currently a lack of targeted generation tools, making it significant to study the theory of personalized reading material generation for international Chinese education and construct corresponding tools.

3 Technical Logic of Personalized Generation of International Chinese Reading Materials

This chapter explores the technical logic of personalized generation of international Chinese reading materials, including learner profiling construction, prompt engineering, and feedback algorithms for reading materials. Learner profiling is based on two indicators: HSK level and learning motivation, which influence the vocabulary, grammar, and types of reading materials. Prompt engineering employs a chain-of-thought prompting strategy to guide large models in generating reading materials that meet teacher expectations, using the SUA API call method. The feedback algorithm draws on the concept of reinforcement learning from human feedback, allowing Chinese teachers to evaluate generated materials and provide feedback to the large model to optimize generation results.

4 Effectiveness Evaluation of Personalized Generation of Reading Materials in International Chinese Education

This chapter focuses on evaluating the effectiveness of large language models in generating reading materials for international Chinese education. The study assesses the quality of generated reading materials from two dimensions—text characteristics and teaching attributes—using a combination of machine and human evaluation methods. Machine evaluation results show that the generated materials are relatively stable in terms of character familiarity and vocabulary diversity, while the total number of sentences and average sentence length exhibit greater fluctuations. In human evaluation, frontline teachers rated the generated materials between 3 and 4 points on average, meeting basic teaching requirements. The study also compares the quality of four generation methods, finding that the abbreviation generation method performs the best and is suitable for learners at all levels; the direct generation method is suitable for advanced learners; imitation writing is suitable for beginners; and expansion writing can serve as an auxiliary method. Additionally, the study analyzes the generation effects under different learning motivations, revealing that generation results for motivations such as "degree acquisition," "written communication," and "external requirements" are unsatisfactory, while results for other motivations are relatively stable. Finally, through questionnaire follow-ups, most frontline teachers believe that reading materials generated by large models have certain practical value and can be used in teaching after appropriate modifications.

5 Resource Innovation in Personalized Generation of Reading Materials for International Chinese Education

This chapter introduces the development and application of a platform for the personalized generation of reading materials for international Chinese education. Based on large language models, the platform features a user-friendly interface where teachers can input relevant parameters to generate reading materials and accompanying exercises that meet HSK level requirements. The platform supports various generation methods, allowing teachers to select according to their needs and input external resources. Generated materials can be downloaded and distributed to students, suitable for pre-class, in-class, and post-class teaching activities, reducing teachers' lesson preparation workload and improving teaching effectiveness. Simultaneously, the platform embodies a "human-AI co-creation" model for teaching resource development, where teachers act as demand proposers and quality supervisors, guiding large models to produce high-quality teaching resources.

6 Conclusion

This study explores the technical pathways for personalized generation of international Chinese reading materials, constructs a "human-AI collaborative co-creation" model, and achieves "teaching students according to their aptitude" and "reducing teachers' workload." Through teacher validation and platform development, the study confirms the effectiveness and operability of large models in reading material generation tasks. However, limitations exist, such as the simplistic modeling of learner profiles, which could be expanded with additional dimensions in the future, and the need to broaden the platform's target user base. The application of large models provides new possibilities for international Chinese education. Future efforts should promote deeper integration with teaching scenarios, assisting teachers in focusing on instructional design and personalized tutoring. The human-AI collaborative co-creation model can also be extended to other teaching resource development areas, driving the digital transformation and high-quality development of international Chinese education.