{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2609.02207", "kind": "paper", "url": "https://arxiv.org/abs/2609.02207"}], "authors": ["Vishnu Prasad Vijaya Kumar", "Santhosh Venkatesh", "Ivan P. Yamshchikov"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2609.02207", "source_url": "https://arxiv.org/abs/2609.02207"}, "categories": ["General AI", "Vision & 3D", "cs.CV"], "description": {"en": "Proposes a challenge set of 500 document images annotated with GDPR-aligned PII to evaluate document-level leakage risk in redaction systems."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2609.02207", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2609.02207", "title": "LeakageBench"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:2609.02207", "modality": null, "name": "LeakageBench", "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": "false", "origin_record_id": "bm_leakagebench_a7e3cbaa", "origin_source": "arxiv", "recognition_confidence": "1.0", "record_sha256": "cbbbbdb8b799a623ec02cd536976093b940031de6824aad6f78335f4f251d8d7", "relation": "introduces", "review_model": "deepseek-v4-pro", "review_state": "ai-name-audit-deferred", "reviewed_at": "2026-09-04T00:46:59.007339Z", "source_url": "https://arxiv.org/abs/2609.02207"}, "publisher": null, "released": "2026-09-02", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2609.02207", "source_url": "https://arxiv.org/abs/2609.02207"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2609-02207", "source": "claire_radar", "source_benchmark_id": "2609.02207", "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": null, "hfPaperUrl": "https://huggingface.co/papers/2609.02207", "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"}, "attentionForecast": {"confidence": "Low", "horizon": "7d", "reason": "Without public artifacts, the benchmark cannot be used or verified, limiting its potential reach.", "score": 0}, "capabilities": [], "confidence": "High", "construction": "Unknown", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-09-04T00:46:59.007339Z", "inputHash": "9ce7117cc26880576df32921ee940adb3026a860f97965111c233d0412f83f83", "model": "deepseek-v4-pro"}, "curation": {"decisionReason": "No code, data, or other artifact links are supplied, making the benchmark's public path and scoring unverifiable.", "model": "deepseek-v4-pro", "policy": "source-evidence-v2", "reviewedAt": "2026-09-04T00:46:59.007339Z", "state": "ai-name-audit-deferred"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "Proposes a challenge set of 500 document images annotated with GDPR-aligned PII to evaluate document-level leakage risk in redaction systems.", "displayEligible": false, "domainCuration": {"method": "rules-v1", "state": "auto"}, "evaluationMode": "public_reusable", "evidence": {"reasonCodes": ["coined title prefix ending in Bench or Benchmark", "benchmark release stated in one sentence", "evaluation protocol evidence"], "snippet": "We introduce LeakageBench, a challenge set of 500 document images with 11,954 GDPR-aligned PII annotations spanning direct identifiers, linkage keys, and contextual re-identification surfaces."}, "familyId": "bmf_6bff381f0e82", "firstSeenAt": "2026-09-03", "heat": null, "id": "bm_leakagebench_a7e3cbaa", "indexedAt": "2026-09-03T00:55:28.123857Z", "industrySectors": [], "links": {"code": null, "data": null, "paper": "https://arxiv.org/abs/2609.02207", "pdf": "https://arxiv.org/pdf/2609.02207", "project": null, "report": "https://arxiv.org/abs/2609.02207"}, "metrics": [], "motivation": "Real-world personally identifiable information (PII) redaction often operates on document images---scans, screenshots, and PDF renderings---where OCR errors, layout structure, and visual noise determine whether sensitive information is actually removed.", "name": "LeakageBench", "oneLine": "Proposes a challenge set of 500 document images annotated with GDPR-aligned PII to evaluate document-level leakage risk in redaction systems.", "paperTitle": "LeakageBench: Document-Level Leakage Risk for Redacting Personally Identifiable Information in Document Images", "primaryDomain": "General AI", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2609.02207", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-09-03T00:55:28.123857Z"}, "publications": [], "ranking": {"90d": {"confidence": "Low", "coverage": 0.0, "growth": {"components": {"githubStars": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}, "hfDatasetDownloads": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}, "hfPaperUpvotes": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}}, "confidence": "Low", "coverage": 0.0, "elapsedDays": null, "historyDate": null, "rank": null, "score": null, "windowDays": 90}, "level": {"components": {"githubStars": {"normalization": "log-max", "normalized": null, "value": null}, "hfDatasetDownloads": {"normalization": "log-max", "normalized": null, "value": null}, "hfPaperUpvotes": {"normalization": "log-max", "normalized": null, "value": null}}, "confidence": "Low", "coverage": 0.0, "rank": null, "score": null}, "method": "attention-ranking-v13", "rank": null, "score": null}}, "readiness": "Paper only", "recognitionConfidence": 1.0, "relation": "introduces", "releasedAt": "2026-09-02", "source": {"authors": ["Vishnu Prasad Vijaya Kumar", "Santhosh Venkatesh", "Ivan P. Yamshchikov"], "categories": ["cs.CV", "cs.CL"], "id": "2609.02207", "title": "LeakageBench: Document-Level Leakage Risk for Redacting Personally Identifiable Information in Document Images", "type": "arxiv", "url": "https://arxiv.org/abs/2609.02207"}, "sourceUpdatedAt": "2026-09-02T07:21:24Z", "topics": ["cs.CV"], "venueAttempts": [], "whyItMatters": "Could support privacy-preserving document processing by measuring leakage beyond direct identifiers, but the lack of public artifacts limits its use."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
