Metraipolis › archive › dc_03debe2b08a7

PaperBanana-Interact: Scientific Diagram Refinement with Multi-Turn Human Feedback — arXiv:2608.30241

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.30241
License (as archived)
CC0-1.0
Archived
2026-10-17T08:07:26.092Z
SHA-256 of the archived text
72c8883bb3746ba4fa838e4a1a22072f08c4b43ca9a4bbaa6a18a77dfd6bcf7a
Stable URL of this page
https://metraipolis.com/documento/dc_03debe2b08a7

How to cite

PaperBanana-Interact: Scientific Diagram Refinement with Multi-Turn Human Feedback — arXiv:2608.30241 — arXiv (en), https://arxiv.org/abs/2608.30241 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_03debe2b08a7 (Metraipolis archive dc_03debe2b08a7, sha256 72c8883bb3746ba4fa838e4a1a22072f08c4b43ca9a4bbaa6a18a77dfd6bcf7a)
BibTeX
@misc{dc03debe2b08a7,
  title = {PaperBanana-Interact: Scientific Diagram Refinement with Multi-Turn Human Feedback — arXiv:2608.30241},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.30241},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_03debe2b08a7; sha256 72c8883bb3746ba4fa838e4a1a22072f08c4b43ca9a4bbaa6a18a77dfd6bcf7a},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.30241. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Recent efforts have aimed to automate scientific diagram generation from paper content (Lin et al., 2026; Zhu et al., 2026a). However, fully satisfying an author's visual and communicative preferences in a single turn is challenging: in our formative user study (N = 14), all participants requested further revisions after viewing an initial draft, and 86% of them rated the refined diagrams as more satisfactory. Despite the clear demand, the multi-turn workflow remains largely underexplored. To bridge this gap, we present MTPaperBananaBench, a benchmark for multi-turn diagram generation containing 292 images annotated with 3,518 user requirements. To reduce expensive human studies and enable scalable benchmarking, we construct a user simulator that, at each turn, identifies unsatisfied requirements and converts k of them into natural language feedback. Evaluating both requirement satisfaction and overall diagram quality reveals two key failure modes shared across baseline multiturn systems: (1) quality drift, where diagram quality progressively declines over turns, and (2) forgetting, where previously implemented features are lost in subsequent turns. To address these issues, we introduce PaperBanana-Interact, a multi-agent system that refines diagrams via an internal critique-and-refine loop. PaperBanana-Interact consistently improves rather than degrades diagram quality across turns, outperforming baselines by 11.9-18.6 points in quality score and reducing forgetting by 3.7-6.2 points.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-01 in cs.CL).
Categorie arXiv: Computation and Language (cs.CL) ; Computer Vision and Pattern Recognition (cs.CV).
Pagina del lavoro: https://arxiv.org/abs/2608.30241 — 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 2026-10-17T07:47:52.790Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).