# ROZA Graphs: Self-Improving Near-Deterministic RAG through Evidence-Centric Feedback

arXiv: https://arxiv.org/abs/2604.07595 · HTML: https://mlsanity.com/paper/2604.07595/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Knowledge- and tool-augmented reasoning, one of 17 directions in LLM reasoning.
- Builds on 20 of its 28 references that are in the map.
- First posted on arXiv 8 Apr 2026.

## Placed in

- LLM reasoning / [Knowledge- and tool-augmented reasoning](https://mlsanity.com/lr/knowledge-and-tool-augmented-reasoning/)
- LLM agents and tool use / [Web, search, and research agents](https://mlsanity.com/agents/web-search-and-research-agents/)
- Synthetic data and self-improvement for LLMs / [Self-evolving search agents](https://mlsanity.com/synth/self-evolving-search-agents/)

## Builds on (most cited first, 20 of 20)

- [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://mlsanity.com/paper/2005.11401/) (2020)
- [ReAct: Synergizing Reasoning and Acting in Language Models](https://mlsanity.com/paper/2210.03629/) (2022)
- [Reflexion: Language Agents with Verbal Reinforcement Learning](https://mlsanity.com/paper/2303.11366/) (2023)
- [HotpotQA: A Dataset for Diverse, Explainable Multi-hop Question Answering](https://mlsanity.com/paper/1809.09600/) (2018)
- [Generative Agents: Interactive Simulacra of Human Behavior](https://mlsanity.com/paper/2304.03442/) (2023)
- [Retrieval-Augmented Generation for Large Language Models: A Survey](https://mlsanity.com/paper/2312.10997/) (2023)
- [Voyager: An Open-Ended Embodied Agent with Large Language Models](https://mlsanity.com/paper/2305.16291/) (2023)
- [Self-RAG: Learning to Retrieve, Generate, and Critique through Self-Reflection](https://mlsanity.com/paper/2310.11511/) (2023)
- [From Local to Global: A Graph RAG Approach to Query-Focused Summarization](https://mlsanity.com/paper/2404.16130/) (2024)
- [Constructing A Multi-hop QA Dataset for Comprehensive Evaluation of Reasoning Steps](https://mlsanity.com/paper/2011.01060/) (2020)
- [MuSiQue: Multihop Questions via Single-hop Question Composition](https://mlsanity.com/paper/2108.00573/) (2021)
- [Cognitive Architectures for Language Agents](https://mlsanity.com/paper/2309.02427/) (2023)
- [Graph Retrieval-Augmented Generation: A Survey](https://mlsanity.com/paper/2408.08921/) (2024)
- [ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory](https://mlsanity.com/paper/2509.25140/) (2025)
- [MEM1: Learning to Synergize Memory and Reasoning for Efficient Long-Horizon Agents](https://mlsanity.com/paper/2506.15841/) (2025)
- [LinearRAG: Linear Graph Retrieval Augmented Generation on Large-scale Corpora](https://mlsanity.com/paper/2510.10114/) (2025)
- [When to use Graphs in RAG: A Comprehensive Analysis for Graph Retrieval-Augmented Generation](https://mlsanity.com/paper/2506.05690/) (2025)
- [Confident and Wrong: Silent Semantic Failures in Coding Agents](https://mlsanity.com/paper/2603.25764/) (2026)
- [Evaluate-as-Action: Self-Evaluated Process Rewards for Retrieval-Augmented Agents](https://mlsanity.com/paper/2603.09203/) (2026)
- [GAM-RAG: Gain-Adaptive Memory for Evolving Retrieval in Retrieval-Augmented Generation](https://mlsanity.com/paper/2603.01783/) (2026)

## Abstract (arXiv)

Language model agents reason from scratch on every query, discarding their chain of thought after each run. The result is lower accuracy and high run-to-run variance. We introduce reasoning graphs, which persist the per-evidence chain of thought as structured edges. Unlike prior memory that retrieves distilled strategies by query similarity, reasoning graphs enable evidence-centric feedback: for every candidate item, the system traverses all incoming evaluation edges across prior runs to surface how that specific item has been judged before. We further introduce retrieval graphs, which feed a planner that prunes consistently-rejected candidates over successive runs. Together they form a ROZA graph: a self-improving feedback loop in which accuracy gains scale with gold-passage reuse (reasoning graph) and efficiency gains scale with candidate-pool overlap (retrieval graph). The base model remains frozen; all gains come from context engineering via graph traversal. We evaluate on MuSiQue and HotpotQA, plus a high-reuse deployment subset. Four findings stand out. (1) Dose-response: accuracy improves monotonically with evidence-profile coverage, reaching +10.6pp over Vanilla RAG at 50%+ coverage on the same questions (47% error reduction, p<0.0001; per-question Spearman ρ=+0.144, p<10-6, n=1,100). (2) Multi-hop scaling: 4-hop accuracy improves by +11.0pp (p=0.0001). (3) Cross-cluster prediction: the cluster-level gain is predicted by gold-passage reuse density (r=0.604, p=0.001, n=26 clusters). (4) High-reuse Pareto dominance: highest or tied-for-highest accuracy alongside 46% lower cost and 46% lower latency. Per-passage decision consistency across repeated runs (N=73 paired probes, K=10 runs each, two model families, three temperatures) rises by +8 to +13pp on a fixed 20-passage context and by +12 to +21pp when the retrieval graph also prunes (all p<0.005).