跳到正文
arXiv cs.AI· Wei Hou, Jingxin Geng, Xin Chen, Zheqing Jiang·· 4 小时前AI 评分34

LMLineage:面向 LLM 智能体的内容溯源与上下文治理系统

LMLineage: Content Lineage and Context Governance for LLM Agents

AI 导读

LMLineage 是一个无需白盒访问或额外推理的带外系统,可将智能体会话的每句输出归因到支撑它的输入片段并给出校准置信度。基于溯源元数据,它实现概念漂移微压缩:检测不再被引用的输入文档并仅压缩这些部分,同时逐字保留用户指令。

正文 · 原文

View PDF HTML (experimental)

Abstract:Interpretability for AI agents should not be confined to understanding the model's internal weights. It must also cover the agent's reasoning and decision-making process over multi-source context. We present LMLineage, an out-of-band system that attributes every output sentence of an agent session to its supporting input spans with calibrated confidences, without requiring white-box access to the model or any additional inference passes. On top of the lineage metadata, we build concept-drift micro compaction: instead of waiting for the context-length limit and summarizing everything, the system detects when previously used input documents have stopped being referenced ("drifted") and surgically compacts only those, preserving user instructions verbatim. We evaluate on a multi-tier corpus: 71 gold annotated sentences from 10 Chinese web-QA cases, plus 90 turns of independently collected multi-turn sessions across three generator models (gpt-4o-mini, qwen3.8-27b, deepseek-v4-flash). Results show that (i) window-level embedding similarity is far better aligned with LLM paraphrasing than whole document or lexical matching; (ii) the greedy-window fast path is both faster and more accurate on average than exhaustive line-matching (a multiple-comparisons bias in the latter inflates web recall at the cost of overall accuracy); (iii) per-turn attribution completes in 1s on a single GPU, remaining stable as context accumulates to 40 turns; and (iv) concept-drift detection is conservative (0% false compaction) with recall bounded by attribution accuracy. We further describe the concept-drift micro compaction design and an interactive incident-tracing/memory-cleaning case study. A companion patent application covers the lineage construction method.
Subjects: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
Cite as: arXiv:2611.22511 [cs.AI]
  (or arXiv:2611.22511v1 [cs.AI] for this version)
  https://doi.org/10.48550/arXiv.2611.22511

arXiv-issued DOI via DataCite (pending registration)

Submission history

From: Wei Hou [view email]
[v1] Tue, 17 Nov 2026 10:05:59 UTC (331 KB)

来源:arXiv cs.AI · arxiv.org