EMIR$^2$: Evolution-Aware Memory with Intent-Guided Multi-Round Retrieval

2026-09-26

AI Summary

This study addresses the limitations of existing memory systems for LLM-based agents in tracking the evolution of historical facts and integrating dispersed evidence. We propose an evolution-aware memory framework centered on a State Evolution Memory Graph (SEMG), which enables dynamic knowledge maintenance through evidence-based semantic state updates and supports evidence tracing under conflicting scenarios. Additionally, an intent-guided multi-turn retrieval mechanism is designed to adaptively acquire relevant evidence. Experimental evaluations on the LoCoMo and MemConflict benchmarks demonstrate that the proposed method significantly enhances long-term memory utilization, conflict resolution, and complex retrieval performance, achieving relative improvements exceeding 12% on several metrics.

Abstract

Long-term memory enables large language model (LLM) agents to leverage historical interactions for future tasks. However, existing memory systems struggle to utilize continuously evolving historical information, as they often rely on static memory representations and single-round retrieval strategies, failing to track factual changes or integrate distributed evidence across long-term interactions. To address these challenges, we propose \textsc{EMIR}$^{2}$, an \textbf{E}volution-Aware \textbf{M}emory framework with \textbf{I}ntent-Guided Multi-\textbf{R}ound \textbf{R}etrieval, enabling LLM agents to maintain evolving historical knowledge and adaptively retrieve relevant evidence. Specifically, \textsc{EMIR}$^{2}$ constructs a State-Evolving Memory Graph (SEMG) that represents long-term memory as evolving knowledge states supported by temporal event trajectories and evidential associations. By maintaining semantic states through evidence-based updates, SEMG preserves historical evolution and enables evidence tracing under complex and conflicting scenarios. Building upon this, we introduce an intent-guided multi-round retrieval mechanism that iteratively identifies missing evidence and expands retrieval based on accumulated information. Experiments on LoCoMo and MemConflict demonstrate that \textsc{EMIR}$^{2}$ improves long-term memory utilization, dynamic and static conflict handling, and complex retrieval performance, achieving relative improvements of more than 12\% in certain categories. These results highlight the effectiveness of jointly modeling memory evolution and adaptive evidence acquisition for long-term agent interactions.

Problem

• Long-term Memory
• LLM Agents
• Memory Evolution
• Multi-Round Retrieval
• Evidence Integration

Innovation

• Evolution-Aware Memory
• State-Evolving Memory Graph
• Intent-Guided Multi-Round Retrieval
• Long-term Memory
• Evidence Tracing