GAM: Hierarchical Graph-based Agentic Memory for LLM Agents

2026-04-14

AI Summary

This work addresses the challenge faced by large language model agents in balancing the acquisition of new information with the stable retention of existing knowledge during extended interactions. To this end, the authors propose a hierarchical graph-structured memory framework that decouples memory encoding from integration: an event evolution graph temporarily captures dialogue flows, and upon detecting semantic shifts, these events are consolidated into a topic-association network. This design enables simultaneous dynamic awareness and long-term consistency. The approach innovatively separates short-term perception from long-term consolidation and introduces a graph-guided, multi-factor retrieval strategy that effectively mitigates interference and enhances contextual precision. Experimental results demonstrate that the method significantly outperforms state-of-the-art baselines on the LoCoMo and LongDialQA benchmarks, achieving notable improvements in both reasoning accuracy and efficiency.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval
• Multiagent Systems: Agent Communication

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Large language models for search
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

To sustain coherent long-term interactions, Large Language Model (LLM) agents must navigate the tension between acquiring new information and retaining prior knowledge. Current unified stream-based memory systems facilitate context updates but remain vulnerable to interference from transient noise. Conversely, discrete structured memory architectures provide robust knowledge retention but often struggle to adapt to evolving narratives. To address this, we propose GAM, a hierarchical Graph-based Agentic Memory framework that explicitly decouples memory encoding from consolidation to effectively resolve the conflict between rapid context perception and stable knowledge retention. By isolating ongoing dialogue in an event progression graph and integrating it into a topic associative network only upon semantic shifts, our approach minimizes interference while preserving long-term consistency. Additionally, we introduce a graph-guided, multi-factor retrieval strategy to enhance context precision. Experiments on LoCoMo and LongDialQA indicate that our method consistently outperforms state-of-the-art baselines in both reasoning accuracy and efficiency.

Problem

• LLM agents
• long-term memory
• knowledge retention
• context interference
• memory architecture

Innovation

• Graph-based Memory
• Hierarchical Memory Architecture
• Memory Consolidation
• LLM Agents
• Multi-factor Retrieval