arXiv cs.CL· Qiyong Zhong, Jiajie Su, Ming Yang, Li Zhang·· 13 小时前AI 评分22
JevRec：面向 LLM 生成式推荐的语言编码与类 Jev 决策建模
JevRec: Linguistic Encoding with Jev-like Decision Modeling for LLM-based Generative Recommendation
AI 导读
JevRec 提出用可变长度词编码物品的生成式推荐方法，为高频物品生成更短 SID、为长尾物品保留更丰富描述，并用独立长度预测头与内容生成头共享交互历史表示。在三个公开数据集上，JevRec 在所有评测推荐指标上均优于对比基线，训练与推理时间低于 TIGER，并降低了长度与内容预测间的梯度冲突。
正文 · 原文
Abstract:Generative recommendation (GR) represents items as semantic identifiers (SIDs) and autoregressively generates the next item's SID from interaction history. However, most methods assign the same SID length to all items despite highly unequal interaction frequencies. This uniform design repeatedly incurs the cost of fine-grained codes for popular items, while shortening codes to reduce this cost also compresses the semantic representations of long-tail items. Inspired by the compact forms of frequent expressions in natural language, we propose JevRec to encode items as variable-length words and generate the next item from interaction context. Linguistic encoding retains frequency-dependent semantic prefixes from shared codebooks, producing shorter SIDs for popular items and richer descriptions for long-tail items. Yet predicting a sequence-level length and generating token-level content impose different optimization requirements that may conflict under a shared output head. Our Jev-like decision modeling component therefore predicts a length distribution with a dedicated head and generates content conditioned on length through a separate head while sharing the interaction-history representation. Experiments on three public datasets show that JevRec outperforms all compared baselines across the evaluated recommendation metrics. Further analyses show lower training and inference time than TIGER and reduced gradient conflict between length and content prediction.
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2610.36937 [cs.CL] |
| (or arXiv:2610.36937v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2610.36937 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Young Zhong [view email]
[v1]
Tue, 27 Oct 2026 09:59:44 UTC (2,376 KB)
来源：arXiv cs.CL · arxiv.org