DiffuReason: Bridging Latent Reasoning and Generative Refinement for Sequential Recommendation

TL;DR

DiffuReason introduces a unified framework combining latent reasoning and diffusion-based refinement for sequential recommendation, improving accuracy by modeling user intent probabilistically and enabling end-to-end training.

Contribution

It proposes a novel 'Think-then-Diffuse' approach that jointly optimizes reasoning and refinement modules using reinforcement learning, addressing noise accumulation and staged training issues.

Findings

Consistently outperforms baseline models on four benchmarks.

Enhances recommendation accuracy through probabilistic intent modeling.

Validated by large-scale industrial online A/B tests.

Abstract

Latent reasoning has emerged as a promising paradigm for sequential recommendation, enabling models to capture complex user intent through multi-step deliberation. Yet existing approaches often rely on deterministic latent chains that accumulate noise and overlook the uncertainty inherent in user intent, and they are typically trained in staged pipelines that hinder joint optimization and exploration. To address these challenges, we propose DiffuReason, a unified "Think-then-Diffuse" framework for sequential recommendation. It integrates multi-step Thinking Tokens for latent reasoning, diffusion-based refinement for denoising intermediate representations, and end-to-end Group Relative Policy Optimization (GRPO) alignment to optimize for ranking performance. In the Think stage, the model generates Thinking Tokens that reason over user history to form an initial intent hypothesis. In the Diffuse stage, rather than treating this hypothesis as the final output, we refine it through a diffusion process that models user intent as a probabilistic distribution, providing iterative denoising against reasoning noise. Finally, GRPO-based reinforcement learning enables the reasoning and refinement modules to co-evolve throughout training, without the constraints of staged optimization. Extensive experiments on four benchmarks demonstrate that DiffuReason consistently improves diverse backbone architectures. Online A/B tests on a large-scale industrial platform further validate its practical effectiveness.