APEX-MEM: Agentic Semi-Structured Memory with Temporal Reasoning for Long-Term Conversational AI

2026-04-15

AI Summary

This work addresses the challenge of maintaining factual consistency in long-term dialogues with large language models, a setting where existing approaches often suffer from noise sensitivity and unstable responses. The authors propose a novel dialogue memory system that integrates an entity-centric attributed graph, an append-only temporally evolving store, and a retrieval-time reasoning agent capable of resolving conflicts. At query time, the system dynamically generates concise, contextually relevant memory summaries. Built upon a domain-agnostic ontology, the framework supports time-aware reasoning and multi-tool collaborative retrieval. Experimental results demonstrate strong performance, achieving 88.88% accuracy on the LOCOMO question-answering task and 86.2% on LongMemEval, significantly outperforming current state-of-the-art conversation-aware methods.

Technology Category

• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval
• Natural Language Processing: Conversational AI/Dialog Systems
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

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

Abstract

Large language models still struggle with reliable long-term conversational memory: simply enlarging context windows or applying naive retrieval often introduces noise and destabilizes responses. We present APEX-MEM, a conversational memory system that combines three key innovations: (1) a property graph which uses domain-agnostic ontology to structure conversations as temporally grounded events in an entity-centric framework, (2) append-only storage that preserves the full temporal evolution of information, and (3) a multi-tool retrieval agent that understands and resolves conflicting or evolving information at query time, producing a compact and contextually relevant memory summary. This retrieval-time resolution preserves the full interaction history while suppressing irrelevant details. APEX-MEM achieves 88.88% accuracy on LOCOMO's Question Answering task and 86.2% on LongMemEval, outperforming state-of-the-art session-aware approaches and demonstrating that structured property graphs enable more temporally coherent long-term conversational reasoning.

Problem

• long-term conversational memory
• temporal reasoning
• memory reliability
• conversational AI
• information conflict

Innovation

• property graph
• temporal reasoning
• append-only memory
• multi-tool retrieval agent
• entity-centric framework