PrismRec：基于频谱分解的偏好流匹配微视频推荐框架
原标题：Preference Flow Matching with Spectral Factorization for Micro-video Recommendation
AI 摘要
国防科技大学、新加坡国立大学和郑州大学的研究者提出 PrismRec 框架，用于微视频推荐。该框架通过频谱语义分解（SSF）将视频帧分解为静态语义和动态因子，并利用上下文校准偏好匹配（CPM）将用户敏感度作为条件注入生成过程。在四个数据集上的实验表明，PrismRec 相比最先进基线最高提升 22.65%，且推理成本和峰值内存最低。
正文节选
Preference Flow Matching with Spectral Factorization for Micro-video Recommendation Abstract Micro-video recommendation aims to infer user preferences from historical interactions and multimodal video content, thereby identifying the next video of interest. However, prevailing methods compress frame sequences into a single holistic representation, entangling the stable visual semantics and the evolving dynamics that jointly shape user preferences. Meanwhile, diffusion- and flow matching-based re