---
title: "HiGram: Hierarchical Graph Memory for LLM Agents with Path-level Localization and Rewrite"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2608.05095v1"
permalink: "https://feed.astralor.com/items/itm_6bea36a35954bc32ed63c9c6"
published: "2026-08-05T17:32:43.000Z"
score: 78
channel: "ai"
tags: ["HiGram","LLM agent","Hierarchical Graph Memory","Path-level Rewrite"]
---

# HiGram: Hierarchical Graph Memory for LLM Agents with Path-level Localization and Rewrite

## AI commentary

> This paper deserves attention from anyone working on agent memory and long-term reasoning. Its value is not just another graph memory, but solving both 'where to update' and 'how to update' together, reducing retrieval costs caused by memory bloat. It is especially useful for those building agent products, as conflicting memories in long-term dialogues are an unavoidable challenge in real deployment.

## At a glance

Agents for long-term reasoning require memory that can be efficiently and effectively updated over time. Existing graph memory methods typically store all memories in a flat graph, where accumulated historical memories can introduce irrelevant context and increase retrieval cost. Moreover, they often update memory units independently, requiring repeated unit-wise rewrites. To address these issues, HiGram proposes an evolving hierarchical graph memory framework with path-level localization and rewriting. It organizes memory into a coarse-to-fine architecture to reduce irrelevant information, uses query- and update-conditioned MicroGraphs to localize support subgraphs and evidence paths, and applies a coordinated rewriting method to jointly revise intra-unit and inter-unit dependencies. Experiments on long-term conversational QA and conflict-aware memory benchmarks show substantial gains in answer quality and token efficiency, as well as improved answer accuracy and evidence selection under dynamic, static, and conditional conflicts.

## Key points

- Proposes a hierarchical graph memory architecture organizing memories into upper-level nodes and MemoryUnits to reduce irrelevant retrieval information
- Designs MicroGraph-based path-level localization to identify support subgraphs and evidence paths before rewriting
- Introduces a coordinated rewriting method that jointly revises intra-unit memory and inter-unit dependencies for valid dependency structure updates
- Demonstrates substantial improvements over baselines in answer quality and token efficiency on long-term conversational QA and conflict-aware benchmarks

## Full text

Agents for long term reasoning require a memory that can be efficiently and effectively updated over time, as new facts and external feedback continue to arrive. Recently, graph memory has been adopted to offer structural organization for multi-hop retrieval and reasoning. However, existing methods store all memories in a flat graph, and accumulated historical memories can introduce irrelevant contexts and increase the cost of evidence selection during retrieval. Moreover, they typically update memory units independently, requiring repeated unit-wise rewrite to cover related changes. To address these issues, we propose HiGram, an evolving hierarchical graph memory framework with path-level localization and rewriting. Specifically, we first propose a hierarchical graph memory, which organizes the memory into coarse-to-fine architecture composed of upper-level nodes and MemoryUnits, thereby reducing the amount of irrelevant information during retrieval. We further propose MicroGraph-based path-level localization, which leverages query and update conditioned MicroGraphs to identify support subgraph and evidence path before rewrite. Finally, we propose a coordinated rewriting method that jointly revises intra-unit memory and inter-unit dependencies, enable valid dependency structures updating in the localized evidence path. Experiments on benchmarks for long-term conversational question answering and conflict-aware memory evaluation demonstrate that our method demonstrate substantial improvements over baselines in answer quality and token efficiency. Besides, our method improves answer accuracy and query-valid evidence selection under dynamic, static, and conditional conflicts.

---

Verify against the original before citing; curated by Astral Feed: https://feed.astralor.com/items/itm_6bea36a35954bc32ed63c9c6