ES-Mem: Event Segmentation-Based Memory for Long-Term Dialogue Agents

2026-01-12

AI Summary

Existing dialogue memory mechanisms often suffer from semantic fragmentation due to rigid memory granularity and struggle with precise contextual localization because flat retrieval fails to leverage discourse structural cues. Inspired by event segmentation theory, this work introduces dynamic event boundary detection into dialogue memory systems for the first time, proposing a hierarchical memory architecture. The approach employs a dynamic event segmentation module to partition long conversations into semantically coherent event units, enabling memory anchoring and hierarchical retrieval grounded in event-boundary semantics. This design effectively balances semantic integrity with fine-grained contextual localization. Empirical results demonstrate consistent performance gains over strong baselines on two memory benchmarks, while the event segmentation module exhibits strong generalization capabilities on dialogue segmentation datasets.

Technology Category

• Natural Language Processing: Discourse, Pragmatics & Argument Mining
• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval
• Cognitive Modeling & Cognitive Systems: Conceptual Inference and Reasoning

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: Personalized, context-aware and across-device search
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Memory is critical for dialogue agents to maintain coherence and enable continuous adaptation in long-term interactions. While existing memory mechanisms offer basic storage and retrieval capabilities, they are hindered by two primary limitations: (1) rigid memory granularity often disrupts semantic integrity, resulting in fragmented and incoherent memory units; (2) prevalent flat retrieval paradigms rely solely on surface-level semantic similarity, neglecting the structural cues of discourse required to navigate and locate specific episodic contexts. To mitigate these limitations, drawing inspiration from Event Segmentation Theory, we propose ES-Mem, a framework incorporating two core components: (1) a dynamic event segmentation module that partitions long-term interactions into semantically coherent events with distinct boundaries; (2) a hierarchical memory architecture that constructs multi-layered memories and leverages boundary semantics to anchor specific episodic memory for precise context localization. Evaluations on two memory benchmarks demonstrate that ES-Mem yields consistent performance gains over baseline methods. Furthermore, the proposed event segmentation module exhibits robust applicability on dialogue segmentation datasets.

Problem

• memory granularity
• semantic integrity
• flat retrieval
• discourse structure
• episodic context

Innovation

• Event Segmentation
• Hierarchical Memory
• Dialogue Agents
• Memory Architecture
• Context Localization