用知识图谱增强风格可控的生成，确保事实不丢失。
Knowledge-Graph-Gated Defactualization for Style-Controllable and Fact-Preserving Generation in Agentic Conversational AI
- 通过知识图谱提取关键实体并临时替换，再还原验证值。
- 相比纯风格控制，实体恢复率显著提升（p<0.0001）。
- 无需微调模型，适合客服等需保真的对话系统。
部署在事实敏感场景（如客户支持）的代理型大语言模型需同时保持事实正确性与可控风格输出。激活导向技术可在不微调模型的前提下实现风格控制，但缺乏区分可验证事实与风格内容的机制，导致语义泄露。本文提出Defactualize-Steer-Rehydrate（DSR）框架，将带类型、显著性加权的知识图谱（KG）与激活导向结合。DSR通过分层正则、NER或词类分类器提取显著实体，生成前用带类型的占位符替换，生成后基于显著性引导的重填充确定性还原真实值。在600个由A2A生成的客户支持案例（共1200次生成）上，对六种不同参数量（1B–13B）的LLaMA系列模型进行评估，并开展专门的KG消融实验。结果表明，相较于仅使用导向的基线，DSR显著提升了验证实体恢复率（Cohen's d=0.225，Bonferroni校正后p=1.0×10⁻⁴），尽管绝对恢复率仍较低；同时有效维持了跨模型家族的风格控制能力。层间可分离性与导向强度诊断揭示了表征级导向与事实基础之间此前未被探索的交互关系。这些结果证明，显式知识工程可在不依赖模型微调的情况下系统性提升生成AI的可信度、可控性与可复现性。代码、缓存的导向向量与评估脚本已公开以支持复现。
原文摘要 · Abstract (English)
Agentic large language models (LLMs) deployed in fact-sensitive applications such as customer support must simultaneously preserve factual correctness and generate responses in a controllable stylistic register. Activation steering enables fine-tuning-free style control by perturbing hidden representations, but it lacks an explicit mechanism for distinguishing verifiable facts from stylistic content, leading to semantic leakage. We address this challenge through \emph{Defactualize-Steer-Rehydrate} (DSR), a knowledge-engineering framework that integrates a typed, salience-weighted knowledge graph (KG) with activation steering. DSR extracts salient entities using a layered regex or NER or lexical-classifier pipeline, replaces them with typed placeholders prior to steering, and deterministically restores verified values through salience-guided rehydration after generation. DSR is evaluated across six LLaMA-family models (1B--13B parameters) on 600 A2A-generated customer-support cases (1,200 generations), with a dedicated KG ablation study. DSR significantly increases verified-entity recovery relative to a steering-only baseline (Cohen's $d=0.225$, $p_{\text{Bonf}}=1.0\times10^{-4}$), though the absolute recovery rate remains modest, while preserving effective style control across diverse model families. Layer-wise separability and steering-strength diagnostics further show previously unexplored interactions between representation-level steering and factual grounding. hese results demonstrate that explicit knowledge engineering can systematically enhance trustworthy, controllable, and reproducible generative AI without requiring model fine-tuning. Code, cached steering vectors, and evaluation scripts are publicly released to support reproducibility.\footnote{https://github.com/Tanmay-IITDSAI/KG-Gated-Defactualization}
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。