Metraipolis › archive › dc_a32fdac449e3

Catching Hallucinated Citations in Video-LLM Question Answering: A Self-Verification Pipeline and Verifier Ablation Study — arXiv:2608.15574

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.15574
License (as archived)
CC0-1.0
Archived
2026-10-18T18:03:48.717Z
SHA-256 of the archived text
e20111d4084eea4b30e0bb07036a27bfe801ac7b1b4470163977a9204af3d6ba
Stable URL of this page
https://metraipolis.com/documento/dc_a32fdac449e3

How to cite

Catching Hallucinated Citations in Video-LLM Question Answering: A Self-Verification Pipeline and Verifier Ablation Study — arXiv:2608.15574 — arXiv (en), https://arxiv.org/abs/2608.15574 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_a32fdac449e3 (Metraipolis archive dc_a32fdac449e3, sha256 e20111d4084eea4b30e0bb07036a27bfe801ac7b1b4470163977a9204af3d6ba)
BibTeX
@misc{dca32fdac449e3,
  title = {Catching Hallucinated Citations in Video-LLM Question Answering: A Self-Verification Pipeline and Verifier Ablation Study — arXiv:2608.15574},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.15574},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_a32fdac449e3; sha256 e20111d4084eea4b30e0bb07036a27bfe801ac7b1b4470163977a9204af3d6ba},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.15574. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Video question answering systems built on vision-language models often produce timestamped claims with high confidence even when unsupported by the cited frame. This deceptive hallucination arises because timestamps imply grounding without ensuring correctness, increasing user trust but not accuracy. We introduce a pipeline that closes this loop. A retrieval-augmented language model drafts answers with per-claim timestamp citations, and each cited frame is independently re-examined before being shown to the user. We compare against a plain baseline and ablate three verification designs, evaluated on both Apple Silicon (MLX) and Google Colab (HF Transformers, CUDA). Directly asking the vision model whether a frame supports a claim fails completely (0% catch rate on 40 claims) due to sycophancy. Blind re-captioning plus a general LLM judge improves results but is unstable, oscillating between 0% and 100% flagged depending on prompt phrasing. Replacing that judge with a small natural language inference model yields a stable, interpretable verifier that catches 79% of fabricated claims on adversarial false-premise questions while leaving true claims untouched. We release the full pipeline, evaluation harness, and implementations for both Apple Silicon and Colab. Code is available at this https URL .

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-18 in cs.AI).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2608.15574 — 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 2026-10-18T16:29:14.027Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-08-18?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).