arXiv:2606.10694v1 Announce Type: new 
Abstract: Large Language Models (LLMs) are increasingly expected to interact with users over long time horizons. However, due to their finite context window, LLMs cannot retain all past interactions, making long-term memory management essential for storing, updating, and retrieving historical information beyond the context limit. Although recent memory systems attempt to address this issue by storing historical information externally, existing approaches suffer from three key limitations: flat text-based memory organizations fail to capture explicit relations among memories, structured memory systems often destructively overwrite evolving facts, and current retrieval mechanisms remain query-agnostic and passive when evidence is incomplete. REAL constructs long-term conversational memory as a temporal and confidence-aware directed property graph, where each atomic fact is represented with entities, relations, valid-time intervals, confidence scores, and exploration intent labels. During memory construction, REAL adopts a non-destructive temporal update strategy that preserves parallel fact versions and their validity intervals, enabling faithful tracking of fact evolution. During retrieval, REAL anchors query-relevant root entities, decouples their exploration intents, and performs semantic evaluator-guided hybrid beam search to extract compact memory subgraphs. It further incorporates counterfactual inference to repair unreliable retrieval states and recover missing memory evidence through implicit logical relations. Comprehensive experiments demonstrate that REAL substantially improves long-term memory performance over flat-text, graph-based, and existing memory baselines, achieving an average improvement of 22.72\%.

تم تقديم إطار جديد يسمى REAL لتحسين إدارة الذاكرة طويلة الأمد لنماذج اللغة الكبيرة (LLMs). يستخدم هذا الإطار رسمًا بيانيًا موجهًا واعيًا بالزمن والثقة لتمثيل الحقائق الذرية، مما يعالج قيود أنظمة الذاكرة الحالية التي تكافح للاحتفاظ بالتفاعلات التاريخية خارج نافذة السياق.

Se ha introducido un nuevo marco llamado REAL para mejorar la gestión de la memoria a largo plazo de los Modelos de Lenguaje Grande (LLMs). Este marco utiliza un grafo de propiedades dirigido, consciente del tiempo y de la confianza, para representar hechos atómicos, abordando las limitaciones de los sistemas de memoria existentes que luchan por retener interacciones históricas más allá de la ventana de contexto.

Un nouveau cadre appelé REAL a été introduit pour améliorer la gestion de la mémoire à long terme des modèles de langage de grande taille (LLMs). Ce cadre utilise un graphe de propriétés dirigé, conscient du temps et de la confiance, pour représenter des faits atomiques, répondant aux limitations des systèmes de mémoire existants qui peinent à conserver les interactions historiques au-delà de la fenêtre de contexte.

A new framework called REAL has been introduced to enhance long-term memory management for Large Language Models (LLMs). This framework utilizes a temporal and confidence-aware directed property graph to represent atomic facts, addressing the limitations of existing memory systems that struggle with retaining historical interactions beyond the context window.