Research on AI Translation Prompt Design for the Aviation Field
面向航空领域的AI翻译提示词设计研究

摘要

随着大语言模型在翻译领域的广泛应用，AI翻译已成为航空专业文本处理的重要辅助工具，但通用翻译提 示词因缺乏领域约束，易出现术语不规范、语体不符、句式混乱等问题。本文以韩礼德语域理论为基础，从语场、 语旨、语式三个维度设计航空领域专用三维优化提示词，并设置通用基础提示词开展对照实验。实验选取航空典型 例句为语料，结合BLEU指标与人工评价开展定量、定性分析。结果显示，相较于基础提示词，三维优化提示词下 所有样本的BLEU值有所提升，能够改善通用提示词在航空翻译中的各类问题。研究表明，将语域理论融入提示词 设计具备可行性，可为航空及同类专业领域AI翻译提示词的研发与应用提供一定参考。

Abstract

With the extensive application of large language models in the field of translation, AI translation has become an important auxiliary tool for processing aviation-related professional texts. However, general translation prompts often lack domain-specific constraints, which can easily lead to problems such as nonstandard terminology, inappropriate style, and disordered sentence structures. Based on Halliday’s register theory, this paper designs three-dimensional optimized prompts for the aviation field from the three dimensions of field, tenor, and mode, and sets up general basic prompts for a comparative experiment. Typical aviation example sentences are selected as the corpus, and quantitative and qualitative analyses are conducted by combining BLEU metrics with manual evaluation. The results show that, compared with basic prompts, the BLEU scores of all samples are improved under the three-dimensional optimized prompts, which can alleviate various problems of general prompts in aviation translation. The study indicates that integrating register theory into prompt design is feasible and can provide reference for the development and application of AI translation prompts in aviation and similar professional fields.