arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~

项目上的动作偏好流:用于预测性和生成性个性化的共享事件模式

Action-On-Item Preference Flow: A Shared Event Schema for Predictive and Generative Personalization

Parthiv Chatterjee, Kashish Kanjaria, Vashisth Purani, Sourish Dasgupta, Tanmoy Chakraborty

arXiv 2610.01375首次发表:更新:

AI 中文总结

本文提出项目上的动作偏好流模式,通过共享更新机制统一处理不同交互历史,并实现预测性与生成性个性化,实验验证了跨源学习的有效性。

AI 中文摘要

用户的电影、新闻和对话历史在其原生动作和输出上各不相同,但每次交互都提供了可以更新用户记忆的证据。我们研究这些历史是否能够训练一个可复用的更新机制。项目上的动作模式将映射的交互角色与内容嵌入配对,允许共享的更新参数在不同的用户状态上操作。我们建立了对原生重新标记的不变性、在项目嵌入扰动下有界的状态变化,以及在显式兼容性条件下的池化训练界限。多时间尺度状态假设(MTSH)规定了这些证据如何进入、持续和被消耗;PerTIDE通过动作门控、三个状态空间轨迹、融合和命令条件读取来实现它。在PENS上,相同的历史编码器同时支持下一篇新闻预测和个性化标题生成。在受控的PENS到MovieLens实验中,冻结的源训练核心在拟合相同目标消费者后,比结构相同的随机核心高出15.23个MRR点。在MIND上,PerTIDE相对于相同输入的三分支状态空间控制保持了4.12点的MRR优势。动作、读取和轨迹干预识别了这些增益的互补贡献。综合来看,理论和实验支持在兼容源之间学习历史更新,并通过预测性和生成性消费者重用它们。

英文摘要

A user's movie, news, and dialogue histories differ in their native actions and outputs, yet each interaction supplies evidence that can update user memory. We study whether these histories can train one reusable update mechanism. An action-on-item schema pairs a mapped interaction role with a content embedding, allowing shared update parameters to operate on separate user states. We establish invariance to native relabeling, bounded state changes under item-embedding perturbations, and a pooled-training bound under explicit compatibility conditions. The Multi-Timescale State Hypothesis (MTSH) specifies how this evidence enters, persists, and is consumed; PerTIDE implements it with action gating, three state-space traces, fusion, and command-conditioned readout. On PENS, the same history encoder supports both next-news prediction and personalized headline generation. In a controlled PENS-to-MovieLens experiment, a frozen source-trained core exceeds an identically structured random core by 15.23 MRR points after fitting the same target consumer. On MIND, PerTIDE retains a 4.12-point MRR advantage over a same-input three-branch state-space control. Action, readout, and trace interventions identify complementary contributions to these gains. Together, the theory and experiments support learning history updates across compatible sources and reusing them through predictive and generative consumers.

CommentsAccepted to NeurIPS 2026. Author-prepared archival version with expanded discussion and interpretation

论文原文

arXiv 摘要页 · PDF 原文 · HTML 原文

↑