Metraipolis › archive › dc_ce9cc7ff92b6
Learning Evidence Sufficiency Boundaries for Selective Answering in Grounded Multi-Hop QA — arXiv:2609.01687
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.01687
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-17T04:16:29.021Z
- SHA-256 of the archived text
50f95f2215b5e16f883419246b08dd0c0a179d90c8e0b027f5aaa5579ce724c0- Stable URL of this page
https://metraipolis.com/documento/dc_ce9cc7ff92b6
How to cite
Learning Evidence Sufficiency Boundaries for Selective Answering in Grounded Multi-Hop QA — arXiv:2609.01687 — arXiv (en), https://arxiv.org/abs/2609.01687 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_ce9cc7ff92b6 (Metraipolis archive dc_ce9cc7ff92b6, sha256 50f95f2215b5e16f883419246b08dd0c0a179d90c8e0b027f5aaa5579ce724c0)
BibTeX
@misc{dcce9cc7ff92b6,
title = {Learning Evidence Sufficiency Boundaries for Selective Answering in Grounded Multi-Hop QA — arXiv:2609.01687},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.01687},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_ce9cc7ff92b6; sha256 50f95f2215b5e16f883419246b08dd0c0a179d90c8e0b027f5aaa5579ce724c0},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.01687. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Grounded question answering systems should answer only when the supplied evidence supports the answer. In multi-hop QA, this requirement is difficult because partial evidence can make an unsupported answer appear plausible. We study selective answering through evidence sufficiency boundaries: for the same question, a model should abstain under unsupported or partially supported context, answer when the context first becomes sufficient, and keep the answer stable when redundant evidence is added. We introduce Evidence Sufficiency Boundary Training, a generation-native training framework that constructs ordered evidence chains and supervises the abstain-to-answer transition directly. The method combines level supervision, a boundary flip margin, post-boundary stability, and answer recall protection. We build evidence chains from HotpotQA, 2WikiMultiHopQA, and MuSiQue, then evaluate models with chain metrics, raw QA utility, and unsupported-answer rates on external non-answerable sets. With Qwen2.5-3B-Instruct and LoRA adaptation, Evidence Sufficiency Boundary Training gives the strongest boundary localization among the tested systems, with flip accuracy of 0.807 compared with 0.781 for a token-level abstention baseline. It also achieves the lowest overall unsupported-answer rate on external non-answerable evaluation, 0.095 compared with 0.101 for the same baseline, while retaining competitive raw QA F1. The results show that grounded selective answering improves when training marks the evidence level where refusal should give way to answering. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-03 in cs.CL). Categorie arXiv: Computation and Language (cs.CL). Pagina del lavoro: https://arxiv.org/abs/2609.01687 — 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 2026-10-17T03:26:49.754Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-03?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).