ClockRoPE: Random Fourier Rotations for Temporal Routine Modeling

AI 中文解读

核心亮点：AI推荐系统现在能像记住“每天几点该干嘛”一样，精准捕捉用户行为的时间规律，让内容推荐更贴合你的日常节奏。通俗解读：过去AI处理文字时用的“位置编码”技术，能帮模型记住词语的先来后到，但遇到用户行为中“每周五晚上看综艺”这种周期性习惯就失灵了。研究人员发现，只要把频率调整成类似钟表指针的周期模式，AI就能学会识别用户行为中的“生物钟”。他们提出的ClockRoPE方法，通过数学变换让推荐模型像掌握日出日落规律一样，轻松预测你何时会对哪类内容感兴趣。实际影响：这项技术已经被部署在大型视频平台的生产系统中。简单说，你刷到的推荐列表会更聪明——工作日中午推轻松短片，周末下午推长电影，甚至能识别你追剧时的“追更节奏”，让推荐像老朋友一样懂你的步调。开发者也能用更低的计算成本实现这些效果，最终你会感觉“这软件越来越懂我了”。

arXiv:2607.26369v1 Announce Type: new Abstract: Rotary Position Embedding (RoPE) has been widely adopted in transformer-based large language models. However, its log-linear frequency schedule, originally designed to produce long-term attention decay, limits its adoption in domains with more complex distance-correlation patterns, such as temporal periodicity in sequential recommendation. We investigate the expressiveness of general query/key rotations and find that any normalized continuous positive-definite attention modulation function can be approximated by random rotations induced by its own Fourier transform, which we term Random Fourier Rotations. Building on this theory, we propose ClockRoPE for routine modeling in sequential recommendation, where rotation frequencies are derived from periodic attention modulation functions. In online A/B tests, ClockRoPE demonstrates consistent improvements in valued engagement metrics, and has been successfully deployed in production-scale generative retrieval system at a major video-sharing platform.