Jiahao Meng, Qize Yang, Yuxuan Wang, Yunfei Chu, Wensi Dai, Jin Xu, Yunhai Tong · 2026 · arXiv (Cornell University)
DOI: 10.48550/arxiv.2610.31158 · cited by 0 works (OpenAlex) · open-access PDF
Wir führen die Video-Centered Deep Research Report Generation ein. Nach einer Video- und Benutzeranfrage identifiziert ein Agent Forschungsfragen, holt externe multimodale Beweise ab und produziert einen interaktiven Bericht, der mit relevanten Video-Momenten verbunden ist. Diese Aufgabe hat einen direkten praktischen Wert, weil sie Benutzern hilft, relevante Momente zu lokalisieren, geteilte Richtungen zu überprüfen, demonstrierte Prozeduren zu reproduzieren und fundierte Entscheidungen aus Video-Inhalten zu treffen. Wir präsentieren einen kompetenzbasierten Workflow und einen OmniDR-Bench für diese Aufgabe. Für die Bewertung schlagen wir einen interaktiven Bericht über Omni Atomic Rubric Trees (ni-Om-ARTs) vor. Jeder Baum enthält drei Dimensionen, 23 g
Machine translation (Meta M2M100); the English original follows.
Deep research agents synthesize open-web evidence into long reports, but current benchmarks do not evaluate research centered on a source video. We introduce Video-Centered Deep Research Report Generation. Given a video and user request, an agent identifies research questions, retrieves external multimodal evidence, and produces an interactive report linked to relevant video moments. This task has direct practical value because it helps users locate relevant moments, verify claims, reproduce demonstrated procedures, and make informed decisions from video content. We present a skill-based workflow and OmniDR-Bench for this task. For evaluation, we propose Omni Atomic Rubric Trees (Omni-ARTs). Each tree contains three dimensions, 23 shared directions, and task-specific atomic binary criteria. The directions test audiovisual grounding, multimodal evidence use, and interactive presentation. Audited Rubric Construction (ARC) builds these trees through independent evidence collection, automated checks, behavioral calibration, and iterative human review. OmniDR-Bench contains 50 Chinese and English tasks across 19 realworld subcategories. It includes 6,414 criteria. Across 18 agent configurations, the best system scores 63.3 out of 100. Current agents still need broader content coverage, stronger causal and logical analysis across modalities, and more accurate citation support. We further analyze models, harnesses, skill configurations, video categories, and research trajectories. The code and data are released at https://github.com/marinero4972/OmniDR-Bench.
Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese
Topics: Rubric, Computer science, Benchmarking, Workflow, Audit, Deep learning
GET https://store.agentexchange.work/gw/papers/verify?title=...&year=... (0.003 USD), fetch this record as JSON with /gw/papers/paper?id=2610.31158, or list what it cites with /gw/papers/references?id=.... Pay per call in USDC (x402) or with a 19 USD card key for 1,000 calls. Catalogue · llms.txtSources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange