SAPO: Step-Aligned Policy Optimization for Reasoning-Based Generative Recommendation

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Zheng, Zaiyi, Min, Guanghui, Zhu, Yaochen, Wu, Liang, Hong, Liangjie, Chen, Chen, Li, Jundong
Natura: Preprint
Pubblicazione: 2026
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866916021242167296
author Zheng, Zaiyi
Min, Guanghui
Zhu, Yaochen
Wu, Liang
Hong, Liangjie
Chen, Chen
Li, Jundong
author_facet Zheng, Zaiyi
Min, Guanghui
Zhu, Yaochen
Wu, Liang
Hong, Liangjie
Chen, Chen
Li, Jundong
contents Generative recommendation treats next-item prediction as autoregressive item-identifier generation. Specifically, items are encoded as semantic identifiers (SIDs), which are short coarse-to-fine token sequences whose early tokens capture broad semantics and later tokens refine them. Recent work augments this paradigm with reasoning traces and optimizes them via reinforcement learning with verifiable rewards, typically outcome-reward algorithm with exact-match feedback on the generated SID. However, in large-catalog recommendation, exact-match feedback on the generated SID only reports whether the final item is correct; when a generated SID mismatches, outcome-reward cannot identify which SID-token prediction caused the mismatch and may penalize matched SID-token positions together with the mismatched position. We identify that the natural unit of credit assignment in this setting is a single reasoning step (one thinking block paired with one SID token). We instantiate this idea in SAPO (Step-Aligned Policy Optimization): rather than broadcasting one advantage to the whole response, SAPO computes a separate group-relative advantage for each reasoning step and applies it only to the corresponding thinking block and SID token. Across three real-world recommendation datasets, SAPO stabilizes reinforcement-learning training and consistently improves over existing generative recommendation baselines, with the largest gains where sparse exact-match feedback makes reasoning-step credit assignment important. Our results suggest that reinforcement-learning objectives for structured generation should mirror the decoder's own decomposition of the output.
format Preprint
id arxiv_https___arxiv_org_abs_2605_17648
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle SAPO: Step-Aligned Policy Optimization for Reasoning-Based Generative Recommendation
Zheng, Zaiyi
Min, Guanghui
Zhu, Yaochen
Wu, Liang
Hong, Liangjie
Chen, Chen
Li, Jundong
Artificial Intelligence
Generative recommendation treats next-item prediction as autoregressive item-identifier generation. Specifically, items are encoded as semantic identifiers (SIDs), which are short coarse-to-fine token sequences whose early tokens capture broad semantics and later tokens refine them. Recent work augments this paradigm with reasoning traces and optimizes them via reinforcement learning with verifiable rewards, typically outcome-reward algorithm with exact-match feedback on the generated SID. However, in large-catalog recommendation, exact-match feedback on the generated SID only reports whether the final item is correct; when a generated SID mismatches, outcome-reward cannot identify which SID-token prediction caused the mismatch and may penalize matched SID-token positions together with the mismatched position. We identify that the natural unit of credit assignment in this setting is a single reasoning step (one thinking block paired with one SID token). We instantiate this idea in SAPO (Step-Aligned Policy Optimization): rather than broadcasting one advantage to the whole response, SAPO computes a separate group-relative advantage for each reasoning step and applies it only to the corresponding thinking block and SID token. Across three real-world recommendation datasets, SAPO stabilizes reinforcement-learning training and consistently improves over existing generative recommendation baselines, with the largest gains where sparse exact-match feedback makes reasoning-step credit assignment important. Our results suggest that reinforcement-learning objectives for structured generation should mirror the decoder's own decomposition of the output.
title SAPO: Step-Aligned Policy Optimization for Reasoning-Based Generative Recommendation
topic Artificial Intelligence
url https://arxiv.org/abs/2605.17648