{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2606.28406", "kind": "paper", "url": "https://arxiv.org/abs/2606.28406"}], "authors": ["Davie Chen"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2606.28406", "source_url": "https://arxiv.org/abs/2606.28406"}, "categories": ["General AI", "Multimodal"], "description": {"en": "SciDraw-Bench is a benchmark for scientific figure generation, with 32 tasks across eight figure types and ten disciplines. Each task pairs a natural-language prompt with a machine-checkable specification, and evaluation uses four dimensions: Text Fidelity, Semantic Correctness, Structural Quality, and Convention Adherence."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2606.28406", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2606.28406", "title": "SciDraw-Bench"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:2606.28406", "modality": null, "name": "SciDraw-Bench", "openness": {"code_license": null, "data_license": null, "evidence": [], "status": "unknown"}, "provenance": {"confidence": "High", "crawl_bundle": "claire_radar_2026-10-14", "crawled_at": "2026-10-14T09:44:51Z", "data_status": "primary-source-indexed", "display_eligible": "unknown", "origin_record_id": "bm_scidraw-bench_755dfcbf", "origin_source": "arxiv", "recognition_confidence": "1.0", "record_sha256": "2c0a7dd8527d6853d323b46e39b91a510cbbc4d463498bb26cc44dd03ebb214e", "relation": "introduces", "review_state": "unreviewed", "source_url": "https://arxiv.org/abs/2606.28406"}, "publisher": null, "released": "2026-06-24", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2606.28406", "source_url": "https://arxiv.org/abs/2606.28406"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2606-28406", "source": "claire_radar", "source_benchmark_id": "2606.28406", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Unknown", "applicationDomains": ["General AI"], "area": "Multimodal", "attention": {"asOf": "2026-10-05", "attemptedAt": "2026-10-04T15:31:48.030916Z", "githubRepo": null, "githubScope": null, "githubStars": null, "hfDailySubmittedAt": null, "hfDataset": null, "hfDatasetDownloads": null, "hfDatasetLikes": null, "hfPaperUpvotes": null, "hfPaperUrl": "https://huggingface.co/papers/2606.28406", "observedAt": "2026-10-04T03:02:42.754940Z", "signalStatus": {"githubStars": {"state": "not_applicable"}, "hfDatasetDownloads": {"state": "not_applicable"}, "hfDatasetLikes": {"state": "not_applicable"}, "hfPaperUpvotes": {"checkedAt": "2026-10-04T15:31:48.030916Z", "state": "unavailable"}}, "status": "unavailable"}, "capabilities": [], "confidence": "High", "construction": "Unknown", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-08-20T15:22:10.044731Z", "inputHash": "dd1e6f68293f117d665d0b51d23ab571b6b6975f15148dc77a58cb2fe14f239a", "model": "deepseek-v4-flash"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "SciDraw-Bench is a benchmark for scientific figure generation, with 32 tasks across eight figure types and ten disciplines. Each task pairs a natural-language prompt with a machine-checkable specification, and evaluation uses four dimensions: Text Fidelity, Semantic Correctness, Structural Quality, and Convention Adherence.", "domainCuration": {"method": "rules-v1", "state": "auto"}, "evidence": {"reasonCodes": ["exact named benchmark artifact released in abstract", "benchmark release stated in one sentence", "evaluation protocol evidence"], "snippet": "We introduce SciDraw-Bench, a benchmark of 32 structured scientific-figure generation tasks spanning eight figure types and ten disciplines, where each task pairs a natural-language prompt with a machine-checkable specification of required labels, relations, components, conventions, and negative constraints."}, "familyId": "bmf_6c84b0b5c404", "firstSeenAt": "2026-08-19", "githubIndex": {"candidateCount": 0, "indexedAt": "2026-09-17T04:40:34.111665+00:00", "repositories": [], "status": "not-found"}, "heat": null, "id": "bm_scidraw-bench_755dfcbf", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": [], "links": {"code": null, "data": null, "hfPaper": "https://huggingface.co/papers/2606.28406", "paper": "https://arxiv.org/abs/2606.28406", "pdf": "https://arxiv.org/pdf/2606.28406", "project": null, "report": "https://arxiv.org/abs/2606.28406"}, "metrics": [], "motivation": "Text-to-image and multimodal generative models are increasingly used to produce scientific figures such as mechanism diagrams, experimental-design schematics, conceptual frameworks, and graphical abstracts.", "name": "SciDraw-Bench", "oneLine": "SciDraw-Bench is a benchmark for scientific figure generation, with 32 tasks across eight figure types and ten disciplines. Each task pairs a natural-language prompt with a machine-checkable specification, and evaluation uses four dimensions: Text Fidelity, Semantic Correctness, Structural Quality, and Convention Adherence.", "paperTitle": "Can AI Draw Science? A Benchmark for Evaluating Scientific Figure Generation by Text-to-Image and Multimodal Models", "primaryDomain": "General AI", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2606.28406", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-10-14T01:46:58.058199Z"}, "publications": [], "publishers": [{"name": "arXiv", "organizationType": "community", "role": "benchmark-publisher", "sourceUrl": "https://arxiv.org/abs/2606.28406"}], "ranking": {}, "readiness": "Paper only", "recognitionConfidence": 1.0, "relation": "introduces", "releasedAt": "2026-06-24", "source": {"authors": ["Davie Chen"], "categories": ["cs.LG", "cs.CV", "cs.GR"], "id": "2606.28406", "title": "Can AI Draw Science? A Benchmark for Evaluating Scientific Figure Generation by Text-to-Image and Multimodal Models", "type": "arxiv", "url": "https://arxiv.org/abs/2606.28406"}, "sourceUpdatedAt": "2026-06-24T14:21:01Z", "topics": ["Multimodal"], "venueAttempts": [], "whyItMatters": "Fills the gap in evaluating scientific figure generation, which requires correct labels and diagrammatic structure beyond natural-image composition. It provides a protocol for measuring usability of generated figures, aiding development of domain-specific generative systems."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
