Metraipolis › archive › dc_5f500aecc4e5

Where Did the Evidence Go? Diagnosing Proof-Set Loss in Long-Term Conversational Memory — arXiv:2611.10170

Source
arXiv (en)
Original
https://arxiv.org/abs/2611.10170
License (as archived)
CC0-1.0
Archived
2026-11-10T18:42:17.136Z
SHA-256 of the archived text
f2047c1caa1a1aa7b3ad585c5b77434b65cbb198391e145a8d8bfb7c9927ddb6
Stable URL of this page
https://metraipolis.com/documento/dc_5f500aecc4e5

How to cite

Where Did the Evidence Go? Diagnosing Proof-Set Loss in Long-Term Conversational Memory — arXiv:2611.10170 — arXiv (en), https://arxiv.org/abs/2611.10170 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_5f500aecc4e5 (Metraipolis archive dc_5f500aecc4e5, sha256 f2047c1caa1a1aa7b3ad585c5b77434b65cbb198391e145a8d8bfb7c9927ddb6)
BibTeX
@misc{dc5f500aecc4e5,
  title = {Where Did the Evidence Go? Diagnosing Proof-Set Loss in Long-Term Conversational Memory — arXiv:2611.10170},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2611.10170},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_5f500aecc4e5; sha256 f2047c1caa1a1aa7b3ad585c5b77434b65cbb198391e145a8d8bfb7c9927ddb6},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.10170. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Long-term conversational question answering can require several separated dialogue turns that support an answer only together. We introduce EviCover, a retrieval approach designed to retain this complementary evidence. It converts questions into observable evidence requirements, checks requirement coverage using retrieved source identifiers, and selects turns jointly. We track annotated support sets - the benchmark-labeled dialogue turns associated with each question - through retrieval, reranking, and final answer-generation context. We evaluate the baseline memory pipeline on generated, human-verified LoCoMo conversations and human-written REALTALK conversations, and examine EviCover on REALTALK's 108 multi-hop questions. Compared with the hybrid lexical/dense baseline, support-turn recall among 60 candidate turns rises from 0.526 to 0.575, and recall in the final ten-turn context rises from 0.349 to 0.379. The evidence funnel identifies separate discovery and selection failures, while paired answer evaluation shows that greater evidence recall does not yet establish a correctness gain. This analysis locates where complementary evidence is lost and distinguishes evidence retrieval from its use in answer generation.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Computation and Language (cs.CL) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2611.10170 — 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 2026-11-10T18:37:37.576Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).