{"slug": "c-2-u-in-rag-compositional-concept-unlearning-in-retrieval-augmented-generation", "title": "C$^2$U in RAG: Compositional Concept Unlearning in Retrieval-Augmented Generation", "summary": "A new arXiv paper (2610.29223v1) introduces Compositional Concept Unlearning (C²U), a query-side method that suppresses unwanted concepts in retrieval-augmented generation while keeping the encoder and document index fixed. Across five datasets, four encoders, and three seeds, the validation-selected C²U configuration reduced pooled binary target Hit@5 by 4.25 percentage points with non-positive control degradation, and on ArXiv four-concept composition it outperformed an averaged-direction comparator after independent validation gates and first-use test evaluation. The method adds only a bounded edited-minus-original delta to the raw query, routing a profile-derived direction through an augmented atom bank, and the authors report it suppresses target density without increasing budgeted control-query exposure.", "body_md": "arXiv:2610.29223v1 Announce Type: new \nAbstract: Retrieval-augmented generation can expose unwanted concepts through retrieved evidence even when the generator is moderated. We study query-side concept suppression with a fixed encoder and document index. Compositional Concept Unlearning (C$^2$U) routes a profile-derived direction through an augmented atom bank, adding only a bounded edited-minus-original delta to the raw query. Across five datasets, four encoders, and three seeds, the validation-selected configuration reduces pooled binary target Hit@5 by 4.25 percentage points with non-positive control degradation and target-exposure changes. Under the evaluated query transforms, it suppresses target density without increasing budgeted control-query exposure. Separately, ArXiv four-concept composition improves over an averaged-direction comparator after independent validation gates and first-use test evaluation.", "url": "https://wpnews.pro/news/c-2-u-in-rag-compositional-concept-unlearning-in-retrieval-augmented-generation", "canonical_source": "https://www.machinebrief.com/news/cdollar2dollaru-in-rag-compositional-concept-unlearning-in-r-ogzs", "published_at": "2026-10-22 04:00:00+00:00", "updated_at": "2026-10-22 04:41:22.776974+00:00", "lang": "en", "topics": ["artificial-intelligence", "machine-learning", "large-language-models", "ai-research", "ai-safety"], "entities": ["arXiv", "Compositional Concept Unlearning", "C²U", "ArXiv"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/c-2-u-in-rag-compositional-concept-unlearning-in-retrieval-augmented-generation", "markdown": "https://wpnews.pro/news/c-2-u-in-rag-compositional-concept-unlearning-in-retrieval-augmented-generation.md", "text": "https://wpnews.pro/news/c-2-u-in-rag-compositional-concept-unlearning-in-retrieval-augmented-generation.txt", "jsonld": "https://wpnews.pro/news/c-2-u-in-rag-compositional-concept-unlearning-in-retrieval-augmented-generation.jsonld"}}