The field of multilingual e-commerce search is moving towards more data-centric approaches, focusing on improving the quality and relevance of search results through innovative data engineering and processing techniques. Researchers are exploring ways to mitigate the challenges of linguistic diversity, noise in user-generated queries, and limited supervision for low-resource languages. The use of large language models (LLMs) is being complemented by systematic data preprocessing, tailored training strategies, and adaptive integration methods to enhance search relevance and efficiency. Noteworthy papers include:

• A Data-Centric Approach to Multilingual E-commerce Product Search, which presents a practical framework for enhancing performance on query-category and query-item relevance tasks.
• REVISION, a novel framework that integrates offline reasoning mining with online decision-making to optimize e-commerce visual search systems.
• Improving E-commerce Search with Category-Aligned Retrieval, which proposes a category-aligned retrieval system to improve search relevance by predicting product categories from user queries.