Identifying translation biases of cognate terms in Chinese textbooks: A GloVe-based method with semantic similarity analysis

With the increasing internationalization of China and the expansion of Chinese language teaching programs, the issue of translation bias in Chinese language textbooks has become increasingly prominent. To improve the accuracy of identifying translation biases in cognate terms, this study proposes a diagnostic method based on Global Vectors for Word Representation (GloVe) word embeddings and semantic similarity. The experimental results indicate that the proposed method achieves a maximum diagnostic accuracy of 0.93, with the best diagnostic accuracy for pronouns reaching 0.95, an improvement of 0.8 compared to traditional methods. Additionally, the proposed method maintains an F1 score ranging from 0.85 to 0.94 across multiple datasets, with the lowest false positive rate. This method effectively improves the translation quality of cognate terms in Chinese language textbooks, enhances the scientific accuracy of semantic expression evaluation, and provides strong support for improving textbook proofreading efficiency and the quality of Chinese language teaching.