MELD: A Protocol for Merging Knowledge Across Distributed Agentic Memories

ID: a67418a4
原文链接: https://arxiv.org/abs/2608.16357
PDF: https://arxiv.org/pdf/2608.16357v1
作者: Lauri Lovén, Jaakko Sauvola, Jukka Riekki, Sasu Tarkoma
日期: 2026-08-17
更新: 2026-08-17
分类: learning
来源类型: paper
标签: agent-memory, knowledge-graph, federation, crdt, protocol, arxiv
质量评分: 4/5
抓取时间: 2026-08-19T12:47:36Z

中文导读

填补联邦式 agent 记忆的协议空白：agent 之间可互调工具却无法共享所知没有协议能对同一事实的两种措辞做合并关联远隔的事实或在不静默丢弃任何一方主张的前提下处理矛盾MELD 以知识图谱本身为运行时模型，每个记忆体通过五结果程序（插入/合并/关联/冲突/拒绝）接纳外来 claim，判据是 scoped claim-key 同一性嵌入相似度与 NLI 裁决，叠加上下文与新鲜度门控；唯一变更状态的对象是可审计可认证的 Patch绑定标准 pub/sub 传输并配 per-claim 状态 CRDT，无协调者即可在分区与有损路由下自愈防 peer 静默改写；检测到的矛盾保留待裁决而非静默消解HotpotQA distractor 上分布式合并不劣于集中式存储召回优于朴素并集，在线存储少约 11%

为什么值得关注

分布式 agent 记忆合并协议：五结果准入 + 可审计 Patch + CRDT，无协调者自愈且存储省 11%

Grounded in the arXiv abstract (published 2026-08-17; categories: cs.DC, cs.AI, cs.MA).

关键信息

论文标题: MELD: A Protocol for Merging Knowledge Across Distributed Agentic Memories
作者: Lauri Lovén, Jaakko Sauvola, Jukka Riekki, Sasu Tarkoma
arXiv: https://arxiv.org/abs/2608.16357
发布时间: 2026-08-17
arXiv 分类: cs.DC, cs.AI, cs.MA
关联标签: agent-memory, knowledge-graph, federation, crdt, protocol, arxiv

English Abstract

Autonomous agents share a transport and can call each other's tools, but they cannot share what they know: no protocol lets two agents' memories reconcile a fact phrased two ways, link related facts held apart, or reconcile contradictory knowledge without silently discarding either claim. We present MELD, a self-managing coherence mechanism for a federation of agent memories whose run-time model is the knowledge graph itself. Each brain admits every incoming claim through a five-outcome procedure (insert, merge, relate, conflict, or reject), decided from three signals (scoped claim-key identity, embedding similarity, and a natural-language-inference verdict) under context and freshness gates, and acting through exactly one auditable, authenticated Patch, the only object that mutates state. A binding onto standard publish/subscribe transport with a per-claim status CRDT keeps sovereign brains coherent in claim status without a coordinator: self-healing after partitions and under lossy routing, and self-protecting against silent rewrite by a peer, under a benign-fault model. MELD does not adjudicate truth; a detected contradiction is preserved for later adjudication, never silently resolved. On HotpotQA distractor, distributed merge is recall-non-inferior to a centralized store under a pre-specified equivalence test and recall-superior to naive union at about 11% less live storage; the merge classifier separates at AUC 0.968 with a 0.013 false-merge rate on adjudicated candidate pairs; the status CRDT reconverges in 30/30 real partition-heal trials where last-writer-wins manages 11/30; and semantic routing delivers about 3x fewer messages at matched recall. We evaluate on a real computing continuum spanning an operator-grade 5G edge, national HPC, and a local tier, with empirically calibrated thresholds.

English Summary

Autonomous agents can share transport and call each other's tools, but no protocol lets two agents' memories reconcile a fact phrased two ways, link related facts, or handle contradictions without silently discarding either claim. MELD is a self-managing coherence mechanism for federated agent memories whose run-time model is the knowledge graph itself: each brain admits incoming claims via a five-outcome procedure (insert, merge, relate, conflict, reject) decided from scoped claim-key identity, embedding similarity, and an NLI verdict under context and freshness gates, acting through exactly one auditable, authenticated Patch. A publish/subscribe binding with per-claim status CRDTs keeps sovereign brains coherent without a coordinator, self-healing after partitions and lossy routing while protecting against silent rewrite by a peer...

Obsidian Notes

内容由 opencli arxiv paper 拉取 arXiv 元数据与摘要生成。
中文导读与价值判断均锚定在条目已有摘要、论文摘要、作者、日期与分类信息上；未补充论文摘要之外的实验细节。
metadata source: opencli