Xuelong Zheng, Xi Chen

Evaluating Generative AI as a Bilingual Dictionary: Performance and Prompt Optimization in Idiom Retrieval

Introduction

Evaluating generative ai as a bilingual dictionary: performance and prompt optimization in idiom retrieval. Evaluate Generative AI's performance as a bilingual dictionary for English idiom retrieval into Chinese. Discover prompt optimization strategies to enhance AI's lexicographical outputs.

Abstract

Generative AI has been widely discussed and utilized for its practicality in generating lexicographic content. However, the parametric knowledge embedded in pre-trained AI models often results in underperforming outputs when the models are tasked with providing authentic linguistic material. To examine the performance of AI in the retrieval and translation of English idioms into Chinese, this study designed two successive experiments using DeepSeek and Gemini 3.1 Pro. The results identify two advantages of AI-generated idiom entries and introduce a practical prompt optimization strategy for fully utilizing the parametric knowledge embedded in pre-trained AI models in a lexicographical context. Keywords: generative AI, prompt engineering, parametric knowledge, English idioms, bilingual lexicography

Full Text

Comments

You need to be logged in to post a comment.