{"record": {"aliases": ["PereStruct: Multimodal Semantic Assembly for Robust Historical Document Parsing"], "artifacts": [{"id": "arxiv:2606.07661", "kind": "paper", "url": "https://arxiv.org/abs/2606.07661"}, {"id": "gh:maksshandybo/perestruct", "kind": "repo", "url": "https://github.com/makSShandybo/PereStruct"}], "authors": ["Maksim Shandybo", "Ivan Bespalov", "Daniil Yefimov", "Marina Kosheleva", "Alexander Loukianov"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2606.07661", "source_url": "https://arxiv.org/abs/2606.07661"}, "categories": ["General AI", "Multimodal"], "description": {"en": "A 93-page benchmark for mapping detected blocks from complex historical newspaper layouts to their corresponding articles."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2606.07661", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2606.07661", "title": "PereStruct benchmark"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:2606.07661", "modality": null, "name": "PereStruct benchmark", "openness": {"code_license": null, "data_license": null, "evidence": [], "status": "unknown"}, "provenance": {"admission_policy_version": "2026-09-09.1", "confidence": "Medium", "crawl_bundle": "claire_radar_2026-10-14", "crawled_at": "2026-09-25T04:40:22Z", "data_status": "primary-source-indexed", "display_eligible": "true", "origin_record_id": "bm_perestruct-multimodal-semantic-assembly-fo_5c5489b3", "origin_source": "arxiv", "recognition_confidence": "0.55", "record_sha256": "999b0db6bbf18983471421ab9a2c62fa5ee79aa22c3e584022e78f69194f7afa", "relation": "introduces", "review_model": "codex-cli-default", "review_state": "ai-reviewed", "reviewed_at": "2026-09-24T20:00:48.581944Z", "source_url": "https://arxiv.org/abs/2606.07661"}, "publisher": null, "released": "2026-06-03", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2606.07661", "source_url": "https://arxiv.org/abs/2606.07661"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2606-07661", "source": "claire_radar", "source_benchmark_id": "2606.07661", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": ["PereStruct: Multimodal Semantic Assembly for Robust Historical Document Parsing"], "annotation": "Unknown", "applicationDomains": ["General AI"], "area": "Multimodal", "attentionForecast": {"confidence": "Low", "horizon": "7d", "reason": "The resource serves a specialized document-analysis community and includes public code, data, and expert-verified annotations.", "score": 2}, "capabilities": [], "confidence": "Medium", "construction": "Transform Existing", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-09-24T20:00:48.581944Z", "inputHash": "9431dc0ea7d47b185fab94b2d05f8ff2c95974628f5927df507021a80f8bb9ba", "model": "codex-cli-default"}, "curation": {"admissionPolicyVersion": "2026-09-09.1", "canonicalNameEvidence": "a curated PereStruct benchmark of 93 pages", "canonicalNameSource": "abstract", "decisionReason": "Publish: the source formally identifies a curated benchmark, supplies expert-verified ground truth, reports a concrete F1 scoring contract for block-to-article mapping, and provides a public code-and-data link.", "evaluationEvidence": {"reuse": "Code and data available at https://github.com/makSShandybo/PereStruct", "scoring": "achieving an F1 score of 0.904 on block-to-article mapping", "task": "expert-verified ground-truth block-to-article mappings"}, "model": "codex-cli-default", "reviewedAt": "2026-09-24T20:00:48.581944Z", "state": "ai-reviewed"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "A 93-page benchmark for mapping detected blocks from complex historical newspaper layouts to their corresponding articles.", "displayEligible": true, "domainCuration": {"method": "rules-v1", "state": "auto"}, "evaluationMode": "public_reusable", "evidence": {"reasonCodes": ["benchmark term in abstract", "benchmark release stated in one sentence", "public artifact URL"], "snippet": "To support reproducibility and advance research in this domain, we release both the training corpus of 599 annotated pages and a curated PereStruct benchmark of 93 pages with expert-verified ground-truth block-to-article mappings."}, "familyId": "bmf_43ebb103e5d8", "firstSeenAt": "2026-08-19", "heat": null, "id": "bm_perestruct-multimodal-semantic-assembly-fo_5c5489b3", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": [], "links": {"code": "https://github.com/makSShandybo/PereStruct", "data": null, "paper": "https://arxiv.org/abs/2606.07661", "pdf": "https://arxiv.org/pdf/2606.07661", "project": null, "report": "https://arxiv.org/abs/2606.07661"}, "metrics": [], "motivation": "Parsing historical documents with complex, non-standard layouts remains a fundamental bottleneck in large-scale archival digitization.", "name": "PereStruct benchmark", "oneLine": "A 93-page benchmark for mapping detected blocks from complex historical newspaper layouts to their corresponding articles.", "paperTitle": "PereStruct: Multimodal Semantic Assembly for Robust Historical Document Parsing", "primaryDomain": "General AI", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2606.07661", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-08-19T11:05:13.395391Z"}, "publications": [], "readiness": "Runnable", "recognitionConfidence": 0.55, "relation": "introduces", "releasedAt": "2026-06-03", "reviewLifecycle": {"state": "pending"}, "source": {"authors": ["Maksim Shandybo", "Ivan Bespalov", "Daniil Yefimov", "Marina Kosheleva", "Alexander Loukianov"], "categories": ["cs.CV", "cs.DL"], "id": "2606.07661", "title": "PereStruct: Multimodal Semantic Assembly for Robust Historical Document Parsing", "type": "arxiv", "url": "https://arxiv.org/abs/2606.07661"}, "sourceUpdatedAt": "2026-06-03T13:10:47Z", "topics": ["Multimodal"], "venueAttempts": [], "whyItMatters": "It supports repeatable evaluation of document reconstruction systems on degraded, irregular archival layouts."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
