用动态知识图谱让大模型持续学习新知识不遗忘。
DYNA : Dynamic Episodic Memory Networks for Augmenting Large Language Models with Temporal Knowledge Graphs in Continuous Learning
- 用带时间戳的事件图谱做外部记忆，可随时更新。
- 相比微调减少7%遗忘率，时间排序准确率提升5%。
- 图结构越紧密，检索效果越好，适合长期学习场景。
大语言模型在引入新知识时容易遗忘或需昂贵重训练。本文提出DYNA，一种轻量级框架，通过将事件作为节点、时间关系作为带时间戳的有向边构建外部可更新的时序知识图谱，作为大模型的外存记忆。查询时，DYNA利用随机游走和中心性度量检索相关节点，并增强大模型响应。在三个时序回忆任务上评估显示，相较于微调，DYNA减少约7%的灾难性遗忘；相比标准RAG，时序排序准确率提升约5%。图谱聚类系数越高，检索性能越好，表明图结构对效果有直接影响。主要贡献包括：(1) 将情景记忆建模为时序知识图谱，(2) 实现无需重训练的大模型增强，(3) 图谱属性可预测检索表现。
原文摘要 · Abstract (English)
Large Language Models (LLMs) struggle to incorporate new knowledge without forgetting or costly retraining. We propose DYNA, a lightweight framework that augments a frozen LLM with a temporal knowledge graph where events are nodes and temporal relations are directed, timestamped edges. The graph serves as an external, updatable memory. At query time, DYNA retrieves relevant nodes via random walks and centrality measures, then augments the LLM's response. Evaluated on three temporal recall tasks, DYNA reduces catastrophic forgetting by ~7% compared to fine-tuning and improves temporal ordering by ~5% over standard RAG. Higher graph clustering coefficients correlate with better retrieval, showing that graph structure matters. Contributions: (1) episodic memory as temporal KG, (2) retraining-free LLM augmentation, (3) graph properties as predictors of retrieval performance.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。