Summary
This blog post details Etsy's innovative use of large language models (LLMs) to enhance the relevance of its search feature. By implementing a Semantic Relevance Evaluation and Enhancement Framework, Etsy combines human judgment with machine learning to better align search results with user intent. The post discusses the framework's structure, including the roles of high-quality data, semantic relevance models, and the integration of these models into real-time search systems. Continuous improvements in search relevance, driven by LLMs, aim to connect buyers with more suitable products while also supporting smaller sellers.