Error Analysis and Optimization Strategies Research on English Translations of Hainan Tourism Texts under the AI Translation Quality Assessment Framework

Abstract

Guided by the theories of Translation Quality Assessment (TQA), Localization, and Tourism Experience, and employing the COMET (Crosslingual Optimized Metrics for Evaluation of Translation) neural translation quality assessment model, error annotation tools, and a user experience questionnaire platform, this study conducts a systematic analysis of the English-language pages on Hainan's official tourism website. The research identifies typical problems in machine translation, including cultural mistranslations, terminological inconsistencies, and grammatical errors. Notably, the mistranslation of culture-specific terms, such as "Junpo Festival" rendered as "Military Hill Festival," significantly diminishes tourist satisfaction. The study constructs a "Translation Error-Tourism Satisfaction" correlation model and proposes a terminology optimization scheme based on user cognition, offering a theoretical basis and practical guidance for enhancing the effectiveness of cross-cultural tourism communication.