AI Paper Atlas

Research map / Large language models / Retrieval and knowledge

From Chaos to Clarity: A Knowledge Graph-Driven Audit Dataset Generation Framework for LLM Unlearning

Weipeng Jiang, Juan Zhai, Shiqing Ma, Ziyan Lei, Xiaofei Xie, Yige Wang, Chao Shen

AAAI 2026 · Large language models › Retrieval and knowledge · approach: Knowledge editing in LLMs · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
audit with LUCID KG-based suite instead of MUSE's 100 questionsunlearned LLaMA2-7B (no unlearning) · MUSE NewsKMCs by ROUGE: 33 -> 4688Improved
audit with LUCID instead of MUSE, entailment criterionLLaMA2-7B on Books · MUSE BooksKMCs by Entailment: 15 -> 38388Improved
audit GA+KLR unlearned model with LUCID instead of MUSEGA+KLR on News · MUSE NewsKMCs by ROUGE: 18 -> 3702Improved
add redundancy removal between forget and retain KGs (ablation)LUCID on News (GA+KLR) · MUSE NewsKMCs by ROUGE: 13298 -> 3702Improved
add redundancy removal (ablation)LUCID on News · MUSE NewsKMCs by Entailment: 52309 -> 21650Improved
add redundancy removal to the audit set (ablation)LUCID on News · MUSE Newsdataset-level ROUGE score: 19.7% to 26.1% inflation without dedupImproved
add redundancy removal to the audit set (ablation)LUCID on News · MUSE Newsdataset-level Entailment score: 32.4% to 35.2% inflation without dedupImproved
redundancy removal on forget KG (ablation)LUCID KG · MUSE NewsKG edges: 24,763 -> 16,912Improved
redundancy removal on MUSE-covered edgesMUSE audit questions · MUSE Bookscovered edges: 2,922 -> 473Hurt
redundancy removal on MUSE-covered edgesMUSE audit questions · MUSE Newscovered edges: 193 -> 102Hurt
apply LUCID to NPO+GDR-unlearned modelNPO+GDR on Books · MUSE BooksKMCs by ROUGE: 0 -> 1435Improved
KG extraction on narrative Books text vs NewsREBEL-based KG extraction · manual assessmentAccuracy of Knowledge Fact: 0.76 -> 0.61Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy