---
title: "LivingRAG: Augmenting Graph RAG with Experience"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2608.25960v1"
permalink: "https://feed.astralor.com/items/itm_b7cde3531fdf1d9e285e2fb5"
published: "2026-08-26T16:21:06.000Z"
score: 70
channel: "ai"
tags: ["LivingRAG","Graph RAG","Retrieval-Augmented Generation","Knowledge Graph","Experience Reuse"]
---

# LivingRAG: Augmenting Graph RAG with Experience

## AI commentary

> This paper turns RAG from one-off isolated queries into a system that continuously accumulates experience, which is especially inspiring for teams working on knowledge-intensive and multi-hop QA. In real business, many questions are repetitive and interrelated; instead of retrieving from scratch every time, sinking verified reasoning into reusable assets may be key to cutting costs.

## At a glance

LivingRAG introduces a writable and reusable experience store for graph-based retrieval-augmented generation (Graph RAG). It decomposes verified reasoning outcomes into graph signals and summaries: graph signals help retrieval locate entities and passages that were useful in earlier related queries, while summaries provide reference reasoning patterns for answer generation. Reusable signals are identified from shared entities, graph neighborhoods, and question templates in online QA streams. Experiments on multi-hop QA benchmarks show that LivingRAG improves accuracy over strong RAG baselines and reduces completion-token usage when relevant prior experience is reused.

## Key points

- LivingRAG adds a writable experience store to a graph-based retrieval backbone, enabling reuse of reasoning experience in later queries.
- Experience is stored as graph signals and summaries, which aid retrieval and answer generation respectively.
- Reusable signals are extracted from shared entities, graph neighborhoods, and question templates to avoid redundant retrieval and reasoning.
- Experiments on multi-hop QA benchmarks show improved accuracy over strong RAG baselines and reduced completion-token usage.

## Full text

Graph-based RAG improves multi-hop question answering by organizing evidence as a knowledge graph. However, most existing RAG systems process each query in isolation and discard useful reasoning from the LLM's response after inference. As a result, later related queries need to retrieve evidence and reason from scratch. We propose LivingRAG, a Graph RAG framework with writable and reusable reasoning experience. LivingRAG adds a writable experience store to a graph-based retrieval backbone, enabling verified experiences to be reused during inference in two ways. Stored graph signals help retrieval find entities and passages that were useful in earlier related queries. Stored summaries provide a reference reasoning pattern for answer generation. We analyze online QA streams and find reusable signals from shared entities, graph neighborhoods, and question templates. Experiments on multi-hop QA benchmarks show that LivingRAG improves accuracy over strong RAG baselines and reduces completion-token use when relevant prior experience is reused.

---

Verify against the original before citing; curated by Astral Feed: https://feed.astralor.com/items/itm_b7cde3531fdf1d9e285e2fb5