RAD-DPO: Robust Adaptive Denoising Direct Preference Optimization for Generative Retrieval in E-commerce

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Chen, Zhiguo, Sun, Guohao, Qiu, Yiming, Yao, Xingzhi, Li, Mingming, Wang, Huimu, Zhang, Yangqi, Wang, Songlin, Xu, Sulong
Natura: Preprint
Pubblicazione: 2026
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866911627551440896
author Chen, Zhiguo
Sun, Guohao
Qiu, Yiming
Yao, Xingzhi
Li, Mingming
Wang, Huimu
Zhang, Yangqi
Wang, Songlin
Xu, Sulong
author_facet Chen, Zhiguo
Sun, Guohao
Qiu, Yiming
Yao, Xingzhi
Li, Mingming
Wang, Huimu
Zhang, Yangqi
Wang, Songlin
Xu, Sulong
contents Generative Retrieval (GR) is rapidly transforming e-commerce search by replacing traditional multi-stage pipelines with the autoregressive decoding of structured Semantic IDs (SIDs). Despite this architectural efficiency, aligning GR models with nuanced, real-world user preferences remains a critical challenge. While Direct Preference Optimization (DPO) offers an efficient alignment solution, its direct application to structured SIDs suffers from three limitations: (i) it penalizes shared hierarchical prefixes, causing gradient conflicts; (ii) it is vulnerable to noisy pseudo-negatives from implicit feedback; and (iii) in multi-label queries with multiple relevant items, it exacerbates a probability "squeezing effect" among valid candidates. To address these issues, we propose RAD-DPO, which introduces token-level gradient detachment to protect prefix structures, similarity-based dynamic reward weighting to mitigate label noise, and a multi-label global contrastive objective integrated with global SFT loss to explicitly expand positive coverage. Extensive offline evaluations and large-scale online A/B testing on JD.com's core search engine demonstrate that RAD-DPO achieves significant improvements in both retrieval precision and training efficiency, proving its robustness for massive industrial deployments.
format Preprint
id arxiv_https___arxiv_org_abs_2602_23964
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle RAD-DPO: Robust Adaptive Denoising Direct Preference Optimization for Generative Retrieval in E-commerce
Chen, Zhiguo
Sun, Guohao
Qiu, Yiming
Yao, Xingzhi
Li, Mingming
Wang, Huimu
Zhang, Yangqi
Wang, Songlin
Xu, Sulong
Information Retrieval
H.3.3
Generative Retrieval (GR) is rapidly transforming e-commerce search by replacing traditional multi-stage pipelines with the autoregressive decoding of structured Semantic IDs (SIDs). Despite this architectural efficiency, aligning GR models with nuanced, real-world user preferences remains a critical challenge. While Direct Preference Optimization (DPO) offers an efficient alignment solution, its direct application to structured SIDs suffers from three limitations: (i) it penalizes shared hierarchical prefixes, causing gradient conflicts; (ii) it is vulnerable to noisy pseudo-negatives from implicit feedback; and (iii) in multi-label queries with multiple relevant items, it exacerbates a probability "squeezing effect" among valid candidates. To address these issues, we propose RAD-DPO, which introduces token-level gradient detachment to protect prefix structures, similarity-based dynamic reward weighting to mitigate label noise, and a multi-label global contrastive objective integrated with global SFT loss to explicitly expand positive coverage. Extensive offline evaluations and large-scale online A/B testing on JD.com's core search engine demonstrate that RAD-DPO achieves significant improvements in both retrieval precision and training efficiency, proving its robustness for massive industrial deployments.
title RAD-DPO: Robust Adaptive Denoising Direct Preference Optimization for Generative Retrieval in E-commerce
topic Information Retrieval
H.3.3
url https://arxiv.org/abs/2602.23964