为AI代理设计专用记忆数据模型，实现长期、有依据的精准回忆。
FluctlightDB: A Memory Model of Data for AI Agents
- 将长期记忆视为独立数据模型，支持线索触发与溯源权重召回。
- 在多个评测中达97%以上召回率，优于现有向量数据库表现。
- 开源轻量引擎，一键安装即可验证，适合研究者复现与对抗测试。
五十年来，数据系统仅回答两个问题：关系模型关注满足谓词的记录，向量模型关注与查询最接近的向量。二者均未针对长会话中的线索驱动、溯源加权召回而设计。本文提出将长期代理记忆视为独立数据模型——具备独特的写入语义（编码、分离、整合、溯源）和读取语义（通过链接记忆图进行线索驱动激活），并推出嵌入式引擎FluctlightDB，通过experience()和activate()实现该契约。我们不主张对Mem0、Zep或HippoRAG等记忆层的创新，而是提供其底层的嵌入式引擎契约。在LoCoMo（官方证据召回指标；10次对话，1,982个黄金片段）上，CHORUS召回率达99.0%（2026年7月内部复现）。在LongMemEval-S（500题，官方session_recall@8）上，检索测试得97.6%（488/500）；端到端问答使用读者/评判栈得97.4%（487/500）——所用协议不同于引用的厂商排行榜数据，仅作参考。在BEIR SciFact（共享MiniLM嵌入，相同测试框架，Recall Fabric开启）上，CHORUS/PRISM在nDCG@10（0.646 vs. 0.645）和Recall@10（0.792 vs. 0.783）上优于Chroma。另报告自研小规模回归测试套件FAMB（重述10例，其他子测试各1例）宏平均100%——内部验证，非同行基准。陌生人可通过pip install "fluctlightdb[native]"及最小连接->experience->activate脚本在1分钟内验证引擎（编译轮子，非仅源码）。测试框架与冻结JSON采用MIT许可证。本文不宣称新神经科学或新Transformer；仅提出数据栈缺失的一层，并发布可复现、可挑战的引擎。
原文摘要 · Abstract (English)
For fifty years, data systems have answered two questions. The relational model asked which records match a predicate; the vector model asked which vectors lie nearest a query. Neither was built for cue-driven, provenance-weighted recall across long sessions. We propose treating long-term agent memory as a distinct data model -- with its own write semantics (encoding, separation, consolidation, provenance) and read semantics (cue-driven activation across a linked memory graph) -- and present FluctlightDB, an embedded engine that implements this contract via experience() and activate(). We make that case carefully, not categorically: we do not claim novelty over Mem0, Zep, or HippoRAG-style memory layers, only an embedded engine contract beneath them. On LoCoMo (official evidence-recall metric; 10 conversations, 1,982 gold spans), CHORUS recalls 99.0% on an internally reproduced July 2026 run. On LongMemEval-S (500 questions, official session_recall@8), our retrieval harness scores 97.6% (488/500); end-to-end QA with our reader/judge stack scores 97.4% (487/500) -- these layers use different protocols than vendor leaderboard figures we cite for context only. On BEIR SciFact (shared MiniLM embeddings, same harness, Recall Fabric on), CHORUS/PRISM edges Chroma on nDCG@10 (0.646 vs. 0.645) and Recall@10 (0.792 vs. 0.783). We also report a small author-designed regression suite (FAMB; paraphrase n=10, other sub-tests n=1) at 100% macro -- internal validation, not peer benchmark. Strangers can verify the engine in under a minute via pip install "fluctlightdb[native]" and a minimal connect() -> experience() -> activate() script (compiled wheel, not source-only). Harnesses and frozen JSON are MIT-licensed. We claim no new neuroscience and no new transformer; we propose a missing layer of the data stack and release an engine others can reproduce and contest.
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。