{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2606.09578", "kind": "paper", "url": "https://arxiv.org/abs/2606.09578"}], "authors": ["Momina Ahsan", "Sarfraz Ahmad", "Ming Shan Hee", "Roy Ka-Wei Lee", "Preslav Nakov"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2606.09578", "source_url": "https://arxiv.org/abs/2606.09578"}, "categories": ["General AI", "Multimodal"], "description": {"en": "TABVERSE is a controlled multimodal benchmark that aligns the same table content across HTML, Markdown, LaTeX, and rendered images, with question category and difficulty tags. It evaluates LLMs and VLMs on question answering, structural understanding, and structure reconstruction."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2606.09578", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2606.09578", "title": "TABVERSE"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "institutions": ["Mohamed bin Zayed University of Artificial Intelligence"], "institutions_reference": {"basis": "openalex_linked_institutions", "doi": "10.48550/arxiv.2606.09578", "retrieved_at": "2026-11-04", "source_key": "claire-radar:2606.09578", "source_url": "https://openalex.org/W7163994637"}, "key": "claire-radar:2606.09578", "modality": null, "name": "TABVERSE", "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_tabverse_d67e32ec", "origin_source": "arxiv", "recognition_confidence": "1.0", "record_sha256": "b253278a42c4aafe6b46ff98618a637b8bc6d1ccc0580ba37faa002e38702f42", "relation": "introduces", "review_state": "unreviewed", "source_url": "https://arxiv.org/abs/2606.09578"}, "publisher": null, "released": "2026-06-08", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2606.09578", "source_url": "https://arxiv.org/abs/2606.09578"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2606-09578", "source": "claire_radar", "source_benchmark_id": "2606.09578", "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": 0, "hfPaperUrl": "https://huggingface.co/papers/2606.09578", "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", "lastSuccessfulDate": "2026-09-25", "state": "stale"}}, "status": "stale"}, "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": "44564f038262799f85352a80ee39a3c6ef3727ff6cada747f12f71fedd6e7ea5", "model": "deepseek-v4-flash"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "TABVERSE is a controlled multimodal benchmark that aligns the same table content across HTML, Markdown, LaTeX, and rendered images, with question category and difficulty tags. It evaluates LLMs and VLMs on question answering, structural understanding, and structure reconstruction.", "domainCuration": {"method": "rules-v1", "state": "auto"}, "evidence": {"reasonCodes": ["exact coined title identity tied to benchmark evidence", "benchmark release stated in one sentence", "evaluation protocol evidence"], "snippet": "We introduce TABVERSE, a controlled multimodal table benchmark that aligns the same table content across multiple structural formats and rendered images, with question category and difficulty tags."}, "familyId": "bmf_4624dae44f51", "firstSeenAt": "2026-08-19", "githubIndex": {"candidateCount": 0, "indexedAt": "2026-09-17T04:40:34.111665+00:00", "repositories": [], "status": "not-found"}, "heat": null, "id": "bm_tabverse_d67e32ec", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": [], "links": {"code": null, "data": null, "hfPaper": "https://huggingface.co/papers/2606.09578", "paper": "https://arxiv.org/abs/2606.09578", "pdf": "https://arxiv.org/pdf/2606.09578", "project": null, "report": "https://arxiv.org/abs/2606.09578"}, "metrics": [], "motivation": "Large Language Models (LLMs) and Vision-Language Models (VLMs) are increasingly evaluated on table reasoning tasks, but the role of table representation remains under-explored.", "name": "TABVERSE", "oneLine": "TABVERSE is a controlled multimodal benchmark that aligns the same table content across HTML, Markdown, LaTeX, and rendered images, with question category and difficulty tags. It evaluates LLMs and VLMs on question answering, structural understanding, and structure reconstruction.", "paperTitle": "TABVERSE: Benchmarking Cross-Format Table Understanding in LLMs and VLMs", "primaryDomain": "General AI", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2606.09578", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-10-14T01:46:58.058199Z"}, "publications": [], "ranking": {}, "readiness": "Paper only", "recognitionConfidence": 1.0, "relation": "introduces", "releasedAt": "2026-06-08", "source": {"authors": ["Momina Ahsan", "Sarfraz Ahmad", "Ming Shan Hee", "Roy Ka-Wei Lee", "Preslav Nakov"], "categories": ["cs.AI", "cs.CL", "cs.IR"], "id": "2606.09578", "title": "TABVERSE: Benchmarking Cross-Format Table Understanding in LLMs and VLMs", "type": "arxiv", "url": "https://arxiv.org/abs/2606.09578"}, "sourceUpdatedAt": "2026-06-08T14:52:46Z", "topics": ["Multimodal"], "venueAttempts": [], "whyItMatters": "Existing table benchmarks conflate content, format, and modality, obscuring the impact of representation choice. TABVERSE enables isolation of representation effects, providing practical guidance for selecting robust table formats for downstream applications and evaluation."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
