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

Excerpt

arXiv:2610.01375v1 Announce Type: new Abstract: 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