AIGQ: Taobao's End-to-End Generative Architecture for E-commerce Query Recommendation
Alibaba researchers propose AIGQ, a hybrid generative framework for pre-search query recommendations on Taobao. It uses list-level fine-tuning, a novel policy optimization algorithm, and a hybrid deployment architecture to overcome traditional limitations.
đź’ˇ
Why it matters
AIGQ represents a significant technical and business impact for Taobao's e-commerce search experience, overcoming limitations of traditional query recommendation systems.
Key Points
- 1AIGQ is an end-to-end generative framework for e-commerce query recommendation
- 2It uses interest-aware list supervised fine-tuning (IL-SFT) to model complex user intent
- 3The policy is further optimized using interest-aware list group relative policy optimization (IL-GRPO)
- 4A hybrid offline-online deployment architecture balances real-time personalization and computational cost
Details
Pre-search query recommendation, known as
Like
Save
Cached
Comments
No comments yet
Be the first to comment