Construction and Analysis of an Intelligent Bilingual Corpus for the International Marketing of Sichuan-Chongqing Specialties

Abstract

In the context of cross-border e-commerce, the core bottleneck restricting the international dissemination of Bashu characteristic products lies in the quality of marketing text translation. Existing research lacks a systematic framework integrating “vertical corpus construction – human-AI collaboration – effectiveness validation.” To address this gap, the present study adopts a “problem-oriented → resource construction → application validation” pathway. The research is structured into three phases. First, corpus construction involves collecting data from authoritative sources in Sichuan and Chongqing covering eight product categories. After cleaning and standardization, translations are generated through a hybrid approach: matching existing authoritative translations or using AI-generated drafts reviewed by TEM-8-certified translators. Bilingual alignment is achieved via a “tool-assisted + manual verification” process, with sampling accuracy exceeding 99%. Second, system development leverages KIMI AI to generate code for the “Bashu Treasure Repository” corpus system, which supports real-time querying and pagination. Third, the study ultimately constructs a corpus comprising 2,956 bilingual texts across 739 products from eight categories and proposes an “AI draft translation → corpus-based refinement” workflow. Theoretically, it lays a foundation for translation studies in the internationalization of regional products. Practically, it offers an integrated solution combining linguistic resources, technical tools, and workflow optimization.