Metraipolis › archive › dc_104b30bd01ac

Reading Between the Chapters: A Graph-Free RAG Rebuilt With AI and Human Feedback — arXiv:2611.00752

Source: arXiv (en)
Original: https://arxiv.org/abs/2611.00752
License (as archived): CC0-1.0
Archived: 2026-11-03T10:21:59.285Z
SHA-256 of the archived text: b7ecf84df22a8ddbe36b8be719828f0c38727dbbdab0e2e5564562033e6f9cfc
Stable URL of this page: https://metraipolis.com/documento/dc_104b30bd01ac

How to cite

Reading Between the Chapters: A Graph-Free RAG Rebuilt With AI and Human Feedback — arXiv:2611.00752 — arXiv (en), https://arxiv.org/abs/2611.00752 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_104b30bd01ac (Metraipolis archive dc_104b30bd01ac, sha256 b7ecf84df22a8ddbe36b8be719828f0c38727dbbdab0e2e5564562033e6f9cfc)

BibTeX

@misc{dc104b30bd01ac,
  title = {Reading Between the Chapters: A Graph-Free RAG Rebuilt With AI and Human Feedback — arXiv:2611.00752},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2611.00752},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_104b30bd01ac; sha256 b7ecf84df22a8ddbe36b8be719828f0c38727dbbdab0e2e5564562033e6f9cfc},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.00752. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Many questions about a book can only be answered by connecting passages chapters apart, such as a cough heard three times across a horse's life. We introduce CrossChapterQA, 1,853 such questions over 14 public-domain books. Each question lists the two to six chapters it needs (spanning half the book at the median) and the relation linking them, cites evidence from each, and has two independent reference answers graded blind. We use it to build LadderRAG, which indexes a book as passages, summaries and book-level cards for about $0.41 and answers a question for under a cent. Its first version failed mostly at retrieval. We redesigned it four times, steered by an LLM judge's verdicts and our analyses of failed answers: a design-level counterpart of RLAIF and RLHF in which feedback changes the architecture, not model weights. We also report the two redesigns that made answers worse. On the 191 questions its first version got wrong, the final version retrieves over twice as many required chapters on held-out books (recall 0.36 to 0.79), and ten annotators rate its answers 6.5/10 on average (764 ratings).

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2611.00752 — 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 2026-11-03T10:19:49.785Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).