{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2606.01132", "kind": "paper", "url": "https://arxiv.org/abs/2606.01132"}], "authors": ["Issa Sugiura", "Shuhei Kurita", "Yusuke Oda", "Naoaki Okazaki"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2606.01132", "source_url": "https://arxiv.org/abs/2606.01132"}, "categories": ["General AI", "Vision & 3D", "cs.CV"], "description": {"en": "HakushoBench is a Japanese chart and table VQA benchmark built from 33 governmental white papers, containing 2,053 images across over 10 image types with manually annotated QA pairs. It evaluates vision-language models on deep holistic understanding of charts and tables."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2606.01132", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2606.01132", "title": "HakushoBench"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:2606.01132", "modality": null, "name": "HakushoBench", "openness": {"code_license": null, "data_license": null, "evidence": [], "status": "unknown"}, "provenance": {"confidence": "Medium", "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_hakushobench_550d3cd1", "origin_source": "arxiv", "recognition_confidence": "0.85", "record_sha256": "a83c1214f5cd93c655cf6af444e9ddfe268e7e0dac1712d92278650fbbd346d4", "relation": "introduces", "review_state": "unreviewed", "source_url": "https://arxiv.org/abs/2606.01132"}, "publisher": null, "released": "2026-05-31", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2606.01132", "source_url": "https://arxiv.org/abs/2606.01132"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2606-01132", "source": "claire_radar", "source_benchmark_id": "2606.01132", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Unknown", "applicationDomains": ["General AI"], "area": "Vision & 3D", "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": 2, "hfPaperUrl": "https://huggingface.co/papers/2606.01132", "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": "2026-09-30", "state": "stale"}}, "status": "stale"}, "capabilities": [], "confidence": "Medium", "construction": "Unknown", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-08-20T15:22:10.044731Z", "inputHash": "378f519992e7e3b1ad147dd8fbcef2ebcecc8c213ba809501addb3b6ae57375c", "model": "deepseek-v4-flash"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "HakushoBench is a Japanese chart and table VQA benchmark built from 33 governmental white papers, containing 2,053 images across over 10 image types with manually annotated QA pairs. It evaluates vision-language models on deep holistic understanding of charts and tables.", "domainCuration": {"method": "rules-v1", "state": "auto"}, "evidence": {"reasonCodes": ["coined title prefix ending in Bench or Benchmark", "benchmark release stated in one sentence"], "snippet": "As a first instantiation, we introduce HakushoBench, a challenging Japanese chart and table VQA benchmark built from 33 governmental white papers."}, "familyId": "bmf_40c99bc08d37", "firstSeenAt": "2026-08-19", "githubIndex": {"candidateCount": 0, "indexedAt": "2026-09-17T04:40:34.111665+00:00", "repositories": [], "status": "not-found"}, "heat": null, "id": "bm_hakushobench_550d3cd1", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": [], "links": {"code": null, "data": null, "hfPaper": "https://huggingface.co/papers/2606.01132", "paper": "https://arxiv.org/abs/2606.01132", "pdf": "https://arxiv.org/pdf/2606.01132", "project": null, "report": "https://arxiv.org/abs/2606.01132"}, "metrics": [], "motivation": "Understanding chart and table images is essential for applying vision-language models (VLMs) to real-world document understanding.", "name": "HakushoBench", "oneLine": "HakushoBench is a Japanese chart and table VQA benchmark built from 33 governmental white papers, containing 2,053 images across over 10 image types with manually annotated QA pairs. It evaluates vision-language models on deep holistic understanding of charts and tables.", "paperTitle": "HakushoBench: A Japanese Chart and Table VQA Benchmark from Governmental White Papers", "primaryDomain": "General AI", "publication": {"evidence": "Accepted to AACL 2026 (Findings)", "evidenceLevel": "author-claim", "evidenceUrl": "https://arxiv.org/abs/2606.01132", "source": "arxiv-comments", "status": "acceptance_claimed", "venue": "AACL 2026 (Findings)", "verifiedAt": "2026-10-14T01:46:58.058199Z"}, "publications": [], "publishers": [{"name": "HakushoBench Team", "organizationType": "academic-lab", "role": "benchmark-publisher", "sourceUrl": "https://arxiv.org/abs/2606.01132"}], "ranking": {}, "readiness": "Paper only", "recognitionConfidence": 0.85, "relation": "introduces", "releasedAt": "2026-05-31", "source": {"authors": ["Issa Sugiura", "Shuhei Kurita", "Yusuke Oda", "Naoaki Okazaki"], "categories": ["cs.CV"], "id": "2606.01132", "title": "HakushoBench: A Japanese Chart and Table VQA Benchmark from Governmental White Papers", "type": "arxiv", "url": "https://arxiv.org/abs/2606.01132"}, "sourceUpdatedAt": "2026-05-31T10:06:47Z", "topics": ["cs.CV"], "venueAttempts": [{"decisionRaw": "Accepted to AACL 2026 (Findings)", "evidence": [{"level": "author-claim", "observedAt": "2026-10-14T01:46:58.058199Z", "rawValue": "Accepted to AACL 2026 (Findings)", "sourceType": "arxiv-comments", "sourceUrl": "https://arxiv.org/abs/2606.01132"}], "reviewStatus": "accepted", "venueName": "AACL 2026 (Findings)"}], "whyItMatters": "HakushoBench addresses the scarcity of non-English benchmarks for chart and table understanding, providing a challenging evaluation for VLMs in Japanese document domains. It reveals a notable performance gap between open-weight and proprietary models, highlighting areas for improvement in multilingual document AI."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
