arXiv cs.CL· Xingyu Tan, Xiaoyang Wang, Qing Liu, Xiwei Xu, Xin Yuan, Liming Zhu, Wenjie Zhang·· 11 小时前AI 评分34
Ctrl-G：在知识图谱增强的 LLM 推理中控制结构与交互暴露
Ctrl-G: Controlling Structural and Interaction Exposure in Knowledge Graph Augmented LLM Reasoning
AI 导读
Ctrl-G 是首个同时控制结构暴露与交互暴露的 KG-RAG 框架，在六个基准中的五个上取得最高准确率，最多超过最强基线 13.8 分。它通过本地构建查询条件化的净化视图并剪枝，限制传给远程 LLM 的图证据；再用分层控制器结合远程规划与本地连通路径探索，并用本地经验记忆复用已验证推理轨迹、门控远程调用以减少重复披露。
正文 · 原文
Abstract:Knowledge graphs (KGs) provide structured evidence for Web question answering with large language models (LLMs), but sending retrieved evidence to a remote LLM can expose sensitive graph information. Existing privacy treatments use entity anonymization and relation-centric abstraction to preserve reasoning cues, but transmitted topology remains visible and disclosure accumulates across repeated calls. Distinctive topology can support re-identification of masked entities (structural exposure), while successive calls accumulate additional paths and exploration traces into a broader view of the KG (interaction exposure). Simply sending less evidence is insufficient because restricted, stepwise views may omit the bridge paths required for multi-hop reasoning. To address this, we propose Ctrl-G, to the best of our knowledge the first KG-RAG framework to jointly control structural and interaction exposure to a remote LLM while supporting accurate multi-hop reasoning. To control structural exposure, Ctrl-G locally constructs query-conditioned sanitized views through structural aggregation and pruning, limiting the graph evidence transmitted to a remote LLM. A hierarchical controller couples remote planning with local connected-path exploration across topic entities, using question indicators and retrieved experience to prune candidates while grounding and verifying retained evidence against the original KG. To control interaction exposure, local experience memory reuses abstract, verified reasoning traces to guide exploration modes and depths and gate remote LLM calls, reducing repeated disclosure across questions. Ctrl-G achieves the highest accuracy among the compared methods on five of six benchmarks, exceeding the strongest baseline by up to 13.8 points across model configurations.
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2601.08739 [cs.CL] |
| (or arXiv:2601.08739v2 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2601.08739 arXiv-issued DOI via DataCite |
Submission history
From: Xingyu Tan [view email]
[v1]
Tue, 13 Jan 2026 17:14:23 UTC (1,434 KB)
[v2]
Fri, 30 Oct 2026 09:58:07 UTC (2,383 KB)
来源：arXiv cs.CL · arxiv.org