Generative retrieval in e-commerce improves by preserving item ID topology
Don't Rewrite the SID Map: Topology Preservation for Generative Retrieval in Dynamic E-Commerce Search
Information RetrievalArtificial Intelligence
Summary
E-commerce sites often add new items all the time, which means they need to update how products are identified and found in their search systems. The authors found that changing these product IDs too much hurts search accuracy, especially if the search system only looks for the best match. They showed that keeping old product IDs unchanged and just adding new ones helps both old and new product searches work better with only light retraining. An online test of their method improved click rates and sales, showing this approach is effective in real use.
What this means in practice
- •For e-commerce platform engineers: Keep item semantic identifiers fixed when updating product catalogs to maintain search accuracy and quickly adapt to new products with warm-start training.
- •For recommendation system developers: Use topology preservation principles to improve retrieval quality as new items are added continuously to dynamic content databases.
Authors
Jianbo Zhu, Xing Fang, Xiaopeng Chu, Mingmin Jin, Jing Wang, Bokang Wang, Jiaqi Ye, Xiang Zhang
Abstract
Generative retrieval assigns each item a discrete semantic identifier (SID) and retrieves items by autoregressively generating SID sequences. In dynamic environments such as e-commerce, new items arrive continuously, requiring periodic updates to the SID-to-Item mapping and model to maintain retrieval timeliness. Each update, however, can invalidate search paths already learned by the retriever, and the quantitative relationship between mapping change and retrieval quality has not been systematically characterized. We study this problem with extensive controlled experiments over 67 days of large-scale e-commerce search logs and obtain three core findings. (1) SID semantic drift causes severe retrieval degradation that may be masked by large beam sizes: mapping updates reduce HitRate by 41.78-62.05% at beam size 1 but only 8.33-10.85% at beam size 1000, as redundant candidate generation conceals the true severity of drift. (2) The fraction of changed SIDs is strictly monotonically related to retrieval quality: as the remapped fraction increases from 0% to 72.75%, HitRate decreases monotonically, dropping up to 42.66% at beam size 1. Full remapping remains substantially worse than the preservation policy under warm-start training, indicating that short-budget continued training alone cannot repair topology disruption. (3) The SID topology-preserving principle makes even the simplest warm-start strategy highly effective: fixing old-item SIDs, inserting new-items, and warm-starting from the previous model improves old-item HitRate by 10% and new-item HitRate by 109%, validated across two non-overlapping update cycles. These findings distill into a concise deployment principle: insert new-items, freeze old-items; warm-start when possible, otherwise freeze; never change SIDs the model has already learned. A three-week online A/B test on TmallAPP shows UCTCVR +0.29% and GMV +1.16%.