arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2605.06331cs.IR
收藏

自回归语义ID生成在生成推荐中的表达性限制

Expressiveness Limits of Autoregressive Semantic ID Generation in Generative Recommendation

University of California, San Diego(加利福尼亚大学圣迭戈分校);Snap Inc.(Snap公司)

Yupeng Hou, Haven Kim, Clark Mingxuan Ju, Eduardo Escoto, Neil Shah, Julian McAuley

AI总结:

本文研究自回归生成推荐模型中语义ID生成结构对模型表达性的影响,提出Latte方法通过引入潜在token改变解码空间,提升NDCG@10性能。

摘要:

生成推荐(GR)模型通过自回归生成离散令牌序列来生成物品,但此过程诱导了结构化的解码空间,其对模型表达性的影响尚不明确。本文发现GR模型生成的物品概率与解码树结构强相关:树中接近的物品在任何用户下获得相似概率,难以区分。理论证明这种结构相关性阻碍GR模型表示传统协同过滤能捕捉的简单模式。为缓解此问题,本文提出Latte,通过在每个语义ID前注入潜在token,将解码空间从单棵树变为多个潜在token条件的树,创建不同路径,缓解概率耦合,使NDCG@10平均提升3.45%。

英文摘要:

Generative recommendation (GR) models generate items by autoregressively producing a sequence of discrete tokens that jointly index the target item. However, this autoregressive generation process also induces a structured decoding space whose impact on model expressiveness remains underexplored. Specifically, token-by-token generation can be viewed as traversing a decoding tree induced by semantic ID tokens, where leaf nodes correspond to candidate items. We observe that the item probabilities produced by GR models are strongly correlated with this tree structure: items that are close in the tree tend to receive similar probabilities for any given user, making it difficult to distinguish among them based on user-specific preferences. We further show theoretically that such structural correlations prevent GR models from representing even simple patterns that can be well captured by conventional collaborative filtering models. To mitigate this issue, we propose Latte, a simple modification that injects a latent token before each semantic ID, reshaping the decoding space from a single tree into multiple latent-token-conditioned trees. This design creates multiple paths with varying tree distances between items, relaxing tree-induced probability coupling and yielding an average of 3.45% relative improvement on NDCG@10. Our code is available at https://github.com/hyp1231/Latte.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑