Abstract
Systems and methods are disclosed for recommendation ranking using a unified, explicitly decomposed user psychological state vector. A coordination layer loads per-user state values including mood, trust, cognitive mode, topic-indexed satiation, and topic-indexed curiosity, constructs explicit cross-state interaction features (including trust×curiosity, mood×satiation, mode×satiation, and trust×mood), and provides these to a ranking model to compute base scores for candidate items. A conflict resolution engine detects contradictory ranking actions implied by different state dimensions and applies a dominance hierarchy with cascading hard and soft constraints, including mood-based content suppression and trust-gated exploration as hard constraints and satiation discounting as a soft constraint. State is updated incrementally after engagement events with bounded per-event change to enforce within-session coherence so consecutive ranking calls evolve gradually rather than oscillating.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Anonymous, "Unified Dynamic User State Vector with Cross-State Conflict Resolution for Recommendation Ranking", Technical Disclosure Commons, ()
https://www.tdcommons.org/dpubs_series/10731