REVISION:Reflective Intent Mining and Online Reasoning Auxiliary for E-commerce Visual Search System Optimization

Taobao has developed REVISION, a new AI framework that combines large language models with traditional e-commerce visual search systems to better understand implicit user intents and reduce no-click search rates. The system uses offline analysis of historical search data and online reasoning to adaptively optimize search results and platform strategies.

• →Taobao's REVISION framework addresses the problem of high no-click rates in visual search by mining implicit user intents from historical data.
• →The system integrates large language models with traditional search infrastructure through offline reasoning and online decision-making components.
• →REVISION-R1-3B model analyzes query images and historical product data to generate optimization plans for the search pipeline.
• →Experimental results show significant improvements in implicit intent mining efficiency and reduction in no-click rates.
• →The framework provides a paradigm for integrating AI models with existing e-commerce search systems for end-to-end optimization.