From Single to Multi-Granularity: Toward Long-Term Memory Association and Selection of Conversational Agents

TL;DR

MemGAS introduces a multi-granularity memory framework for conversational agents, improving long-term memory retention and response personalization by adaptive selection and filtering of memory units, outperforming existing methods.

Contribution

The paper proposes MemGAS, a novel multi-granularity memory system with adaptive selection and filtering, enhancing long-term dialogue memory in conversational agents.

Findings

Outperforms state-of-the-art methods on four benchmarks.

Effective in diverse query types and retrieval settings.

Enhances coherence and personalization in long-term conversations.

Abstract

Large Language Models (LLMs) have recently been widely adopted in conversational agents. However, the increasingly long interactions between users and agents accumulate extensive dialogue records, making it difficult for LLMs with limited context windows to maintain a coherent long-term dialogue memory and deliver personalized responses. While retrieval-augmented memory systems have emerged to address this issue, existing methods often depend on single-granularity memory segmentation and retrieval. This approach falls short in capturing deep memory connections, leading to partial retrieval of useful information or substantial noise, resulting in suboptimal performance. To tackle these limits, we propose MemGAS, a framework that enhances memory consolidation by constructing multi-granularity association, adaptive selection, and retrieval. MemGAS is based on multi-granularity memory units and employs Gaussian Mixture Models to cluster and associate new memories with historical ones. An entropy-based router adaptively selects optimal granularity by evaluating query relevance distributions and balancing information completeness and noise. Retrieved memories are further refined via LLM-based filtering. Experiments on four long-term memory benchmarks demonstrate that MemGAS outperforms state-of-the-art methods on both question answer and retrieval tasks, achieving superior performance across different query types and top-K settings. \footnote{https://github.com/quqxui/MemGAS}