Graph-based Agent Memory: Taxonomy, Techniques, and Applications

TL;DR

This work surveys graph-based memory for LLM-driven agents, addressing limitations in static knowledge, tool use, and long-horizon tasks. It systematizes memory into knowledge and experience, then details a lifecycle of memory extraction, storage, retrieval, and evolution, with a focus on graph structures such as knowledge graphs, hierarchical graphs, temporal graphs, and hypergraphs, including hybrid architectures. It also covers retrieval operators, enhancement strategies, and memory evolution through internal consolidation and external self-exploration, complemented by open-source libraries, benchmarks, and diverse applications. The paper highlights practical considerations for building scalable, interpretable, and privacy-preserving graph-based memory systems, and maps future directions in dynamic schemas, multi-agent memory coordination, and theory-grounded evaluation.

Abstract

Memory emerges as the core module in the Large Language Model (LLM)-based agents for long-horizon complex tasks (e.g., multi-turn dialogue, game playing, scientific discovery), where memory can enable knowledge accumulation, iterative reasoning and self-evolution. Among diverse paradigms, graph stands out as a powerful structure for agent memory due to the intrinsic capabilities to model relational dependencies, organize hierarchical information, and support efficient retrieval. This survey presents a comprehensive review of agent memory from the graph-based perspective. First, we introduce a taxonomy of agent memory, including short-term vs. long-term memory, knowledge vs. experience memory, non-structural vs. structural memory, with an implementation view of graph-based memory. Second, according to the life cycle of agent memory, we systematically analyze the key techniques in graph-based agent memory, covering memory extraction for transforming the data into the contents, storage for organizing the data efficiently, retrieval for retrieving the relevant contents from memory to support reasoning, and evolution for updating the contents in the memory. Third, we summarize the open-sourced libraries and benchmarks that support the development and evaluation of self-evolving agent memory. We also explore diverse application scenarios. Finally, we identify critical challenges and future research directions. This survey aims to offer actionable insights to advance the development of more efficient and reliable graph-based agent memory systems. All the related resources, including research papers, open-source data, and projects, are collected for the community in https://github.com/DEEP-PolyU/Awesome-GraphMemory.

Figures (8)

• Figure 1: A diagram illustrating the workflow of an AI agent system, and the detailed implementations of the memory system.
• Figure 2: Two types of agent memory, i.e., Knowledge Memory and Experience Memory, and their application across different agent scenarios. The synergy between static knowledge memory and dynamic experience memory enables agents to both understand the world's rules and adapt to personal interactions.
• Figure 3: Comparison between traditional agent memory and graph-based agent memory
• Figure 4: A Comprehensive Taxonomy of Graph-based Memory Management for LLM Agents.
• Figure 5: Overview of agent memory extraction. This figure illustrates a unified pipeline of agent memory construction from various data resources. Raw inputs, originating from both experience and knowledge memory, are transformed into structured and compact representations through specialized extraction techniques. These extracted units are then organized into distinct functional memory types, enabling agents to support reasoning and downstream tasks.

Theorems & Definitions (3)

• Definition 3.1: AI Agents
• Definition 3.2: Basic Operations of Memory
• Definition 3.3: Lifecycle of Agent Memory