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

Abstract

Based on the theories of Translation Quality Assessment (TQA) and localisation and tourism experience, this paper will use the COMET (Crosslingual Optimized Metrics for Evaluation of Translation) neural machine translation quality assessment model, error annotation tools and a user experience questionnaire platform to carry out all-around analysis of the English-language pages on the official tourism website of Hainan Province. The causes of problems in machine translation are generally cultural differences, inconsistencies in terms and grammar, etc. One is that the mistranslation of the culture-specific term "Junpo Festival" has led to a drop in tourist satisfaction. According to the perceptions of users, this paper will build a correlation model for "Translation Error - Tourism Satisfaction" and put forward a terminology optimisation plan to promote the realisation of cross-cultural tourism communication and provide theoretical support.