用扩散模型生成商品编码，解决传统推荐模型顺序依赖和数字重要性不均的问题。
DiffGRM: Diffusion-based Generative Recommendation Model
- 改用掩码扩散模型并行生成商品编码，打破顺序依赖限制。
- 在多个数据集上提升推荐效果，NDCG@10最高增益达15.5%。
- 适合追求高精度与多样性的推荐系统研究者使用。
生成式推荐（GR）将每个物品表示为由n个语义位组成的编号（SID），并通过自回归方式基于用户历史预测下一个物品的SID。然而，SID的两个结构特性使自回归模型（ARM）表现不佳：第一，项内一致性——n个数字共同指定一个物品，但自回归训练仅依赖前缀信息，阻碍了跨数字的双向上下文；第二，数字间异质性——不同数字的语义粒度和可预测性差异大，而统一的下一个词目标对所有数字赋予相同权重，导致易数字过拟合、难数字欠拟合。为此，我们提出DiffGRM，一种基于扩散模型的生成式推荐模型，以掩码离散扩散模型（MDM）替代自回归解码器，实现双向上下文建模与任意顺序并行生成。具体地，我们在三方面进行优化：（1）采用并行语义编码（PSE）解耦数字，平衡各数字信息量；（2）通过策略一致的噪声注入（OCN）优先关注不确定数字，聚焦高价值信号；（3）推理时采用置信度引导的并行去噪（CPD），优先填充高置信度数字，生成多样化的Top-K候选。实验表明，DiffGRM在多个数据集上持续优于强基线，NDCG@10提升6.9%-15.5%。代码已开源。
原文摘要 · Abstract (English)
Generative recommendation (GR) is an emerging paradigm that represents each item via a tokenizer as an n-digit semantic ID (SID) and predicts the next item by autoregressively generating its SID conditioned on the user's history. However, two structural properties of SIDs make ARMs ill-suited. First, intra-item consistency: the n digits jointly specify one item, yet the left-to-right causality trains each digit only under its prefix and blocks bidirectional cross-digit evidence, collapsing supervision to a single causal path. Second, inter-digit heterogeneity: digits differ in semantic granularity and predictability, while the uniform next-token objective assigns equal weight to all digits, overtraining easy digits and undertraining hard digits. To address these two issues, we propose DiffGRM, a diffusion-based GR model that replaces the autoregressive decoder with a masked discrete diffusion model (MDM), thereby enabling bidirectional context and any-order parallel generation of SID digits for recommendation. Specifically, we tailor DiffGRM in three aspects: (1) tokenization with Parallel Semantic Encoding (PSE) to decouple digits and balance per-digit information; (2) training with On-policy Coherent Noising (OCN) that prioritizes uncertain digits via coherent masking to concentrate supervision on high-value signals; and (3) inference with Confidence-guided Parallel Denoising (CPD) that fills higher-confidence digits first and generates diverse Top-K candidates. Experiments show consistent gains over strong generative and discriminative recommendation baselines on multiple datasets, improving NDCG@10 by 6.9%-15.5%. Code is available at https://github.com/liuzhao09/DiffGRM.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。