Memory Matters More: Event-Centric Memory as a Logic Map for Agent Searching and Reasoning

TL;DR

CompassMem proposes an event-centric memory framework that builds a structured Event Graph to explicitly encode logical relations among memory events, inspired by Event Segmentation Theory. Memory construction proceeds incrementally via event segmentation, relation extraction, and graph updates, while memory retrieval uses a Planner–Explorer–Responder team to perform logic-guided, multi-path search over the graph. Empirical results on LoCoMo and NarrativeQA show consistent improvements across backbones, especially for multi-hop and temporal reasoning, demonstrating that organizing memory as a navigable logic map enhances long-horizon reasoning. The approach highlights the importance of structured, relation-aware memory for active reasoning in agents and suggests avenues for scaling to diverse tasks with efficient search.

Abstract

Large language models (LLMs) are increasingly deployed as intelligent agents that reason, plan, and interact with their environments. To effectively scale to long-horizon scenarios, a key capability for such agents is a memory mechanism that can retain, organize, and retrieve past experiences to support downstream decision-making. However, most existing approaches organize and store memories in a flat manner and rely on simple similarity-based retrieval techniques. Even when structured memory is introduced, existing methods often struggle to explicitly capture the logical relationships among experiences or memory units. Moreover, memory access is largely detached from the constructed structure and still depends on shallow semantic retrieval, preventing agents from reasoning logically over long-horizon dependencies. In this work, we propose CompassMem, an event-centric memory framework inspired by Event Segmentation Theory. CompassMem organizes memory as an Event Graph by incrementally segmenting experiences into events and linking them through explicit logical relations. This graph serves as a logic map, enabling agents to perform structured and goal-directed navigation over memory beyond superficial retrieval, progressively gathering valuable memories to support long-horizon reasoning. Experiments on LoCoMo and NarrativeQA demonstrate that CompassMem consistently improves both retrieval and reasoning performance across multiple backbone models.

Figures (6)

• Figure 1: Comparison among CompassMem and the traditional agent memory framework.
• Figure 2: Overview of the proposed CompassMem framework, which contains mainly two part: Incremental Hierarchical Memory Construction and Active Multi-Path Memory Search
• Figure 3: Efficiency–performance trade-off across memory frameworks. Scatter plots compare F1 with construction time, total processing time, per-question latency, and token consumption.
• Figure 4: Ablation results on LoCoMo.
• Figure 5: Scaling results comparing fixed high-capacity and scale-consistent memory construction. Shaded areas show gains from stronger construction models.