论文提出 Memory Echo 现象与 FPA 框架，修复智能体长期记忆失真
Memory Echo in Single-Agent Long-Term Memory: Lifecycle Evaluation and Provenance-Guided Mitigation
做智能体长期记忆的可以看看：论文发现同源旧记录会重复刷存在感，FPA 框架实测把准确率拉高近 10 个点，代码开源可复现。
arXiv 论文定义了 Memory Echo 现象：同源历史记录经重复检索和写回后占据多个检索位置，在没有独立证据的情况下影响当前回答。作者提出 MEchoBench，在 190 个固定问题上评估该现象的出现、纠错阻力和持续性。配套的 Family–Provenance–Authority（FPA）框架让每个来源家族只保留一条代表记录，并以经过验证的修正和权威证据约束最终回答。在四种记忆框架的配对回放中，FPA 将 DeepSeek V4 Flash 的准确率从 0.825 提升到 0.920，Kimi K2.6 从 0.782 提升到 0.916，平均决策污染效应分别从 0.106 和 0.109 降到 0.030 和 0.038。代码已在 GitHub 开源。
Memory Echo in Single-Agent Long-Term Memory: Lifecycle Evaluation and Provenance-Guided Mitigation
Long-term memory supports agents across sessions. Repeated retrieval and write-back can let one historical claim occupy multiple retrieval positions and influence current answers without independent evidence. Existing work examines retrieval, updating, and question answering, leaving same-source record accumulation and post-correction persistence underexplored. Motivated by research in social cognition and memory psychology on repeated exposure, retrieval, and revision, we define \emph{Memory Echo} as the excess retrieval exposure gained by historical information from the same source without additional independent evidence, and introduce MEchoBench to evaluate its emergence, correction resistance, persistence, and effect on answers. To address this phenomenon, we introduce the Family--Provenance--Authority framework (FPA). Specifically, FPA selects one representative per source family, conditions the eligibility of old records for the answer context on verified corrections, and constrains final answers with authoritative evidence. In paired replay on 190 fixed questions across four memory frameworks, FPA raises accuracy from 0.825 to 0.920 with DeepSeek V4 Flash and from 0.782 to 0.916 with Kimi K2.6; mean Decision Contamination Effect falls from 0.106 to 0.030 and from 0.109 to 0.038, respectively. These results show that agent memory can preserve traceable history without allowing repeated historical information to govern current decisions. The code is available at https://github.com/OrcasSam/memory-echo-benchmark.
