Metraipolis › archive › dc_7943e5bdefff

Conformal Factuality Control for Multi-Hop Retrieval-Augmented Generation — arXiv:2609.38222

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.38222
License (as archived)
CC0-1.0
Archived
2026-10-07T13:13:18.914Z
SHA-256 of the archived text
6b0540f30a11de6971bd3e9db39148695af5a6274afb566d00f9ea9f86a4e278
Stable URL of this page
https://metraipolis.com/documento/dc_7943e5bdefff

How to cite

Conformal Factuality Control for Multi-Hop Retrieval-Augmented Generation — arXiv:2609.38222 — arXiv (en), https://arxiv.org/abs/2609.38222 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_7943e5bdefff (Metraipolis archive dc_7943e5bdefff, sha256 6b0540f30a11de6971bd3e9db39148695af5a6274afb566d00f9ea9f86a4e278)
BibTeX
@misc{dc7943e5bdefff,
  title = {Conformal Factuality Control for Multi-Hop Retrieval-Augmented Generation — arXiv:2609.38222},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.38222},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_7943e5bdefff; sha256 6b0540f30a11de6971bd3e9db39148695af5a6274afb566d00f9ea9f86a4e278},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.38222. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Retrieval-augmented generation (RAG) can ground large language models in external evidence, but retrieved context does not guarantee that generated claims are factually supported. This problem is especially relevant in multi-hop RAG, where retrieval and reasoning proceed through multiple dependent stages. We study whether claim-level conformal factuality control, previously developed for RAG, remains effective in this setting. We apply split-conformal claim filtering to multi-hop RAG and evaluate it on HotpotQA, Natural Questions, and TriviaQA using Llama 3.1 8B and GPT-4o-mini, together with a single-hop reference experiment. Across all six multi-hop model-dataset configurations, increasingly stringent conformal targets consistently increase the fraction of responses whose retained claims are fully supported. At the 95% target, this rate ranges from 95.80% to 97.20%, compared with 55.60%-76.03% without filtering. However, the improvement is strongly selective: only 4.41%-31.09% of generated claims are retained and 9.70%-51.40% of responses remain non-empty at the 95% target. These results show that conformal factuality extends to multi-hop RAG, while demonstrating that nominal reliability must be interpreted jointly with claim retention and abstention.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-10-01 in cs.CL).
Categorie arXiv: Computation and Language (cs.CL) ; Machine Learning (cs.LG); Machine Learning (stat.ML).
Pagina del lavoro: https://arxiv.org/abs/2609.38222 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-07T12:44:51.712Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-10-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).