OTR — Observation-to-Relationship: Persistent Relationships for Traceable AI Reasoning
Year: 2026
Type: article
Abstract: OTR (Observation-to-Relationship) introduces a model-agnostic, append-only evidentiary architecture for persistent AI reasoning. It preserves observations, relationships, and reviews across inference events to support provenance tracking, contradiction resolution, and longitudinal analysis. The paper presents a reference implementation and experimental evaluations spanning scientific retractions, factual validity, cross-domain perception, and computational scaling, accompanied by an interactive explorer and reproducibility tools.
Author: Shapira, Leeor
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7233760304
