{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2605.29738", "kind": "paper", "url": "https://arxiv.org/abs/2605.29738"}, {"id": "hf:overthelex/multi-legal-bench", "kind": "dataset", "url": "https://huggingface.co/datasets/overthelex/multi-legal-bench"}], "authors": ["Volodymyr Ovcharov"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2605.29738", "source_url": "https://arxiv.org/abs/2605.29738"}, "categories": ["Language & Knowledge", "Reasoning", "Software & AI Compute"], "description": {"en": "Multi-Legal-Bench evaluates LLMs on legal reasoning across six countries, four language families, and five tasks, using structured metadata from court registries for classification and extraction."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2605.29738", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2605.29738", "title": "Multi-Legal-Bench"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:2605.29738", "modality": null, "name": "Multi-Legal-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_multi-legal-bench_2cfb13a9", "origin_source": "arxiv", "recognition_confidence": "1.0", "record_sha256": "c24957daf9707620084c401be706e9e552315dc4a33120bd9310c24eabdbe0dd", "relation": "introduces", "review_state": "unreviewed", "source_url": "https://arxiv.org/abs/2605.29738"}, "publisher": null, "released": "2026-05-28", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2605.29738", "source_url": "https://arxiv.org/abs/2605.29738"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2605-29738", "source": "claire_radar", "source_benchmark_id": "2605.29738", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Mixed", "applicationDomains": ["Software & AI Compute"], "area": "Language & Knowledge", "attention": {"asOf": "2026-10-05", "attemptedAt": "2026-10-04T15:31:48.030916Z", "githubRepo": null, "githubScope": null, "githubStars": null, "hfDailySubmittedAt": null, "hfDataset": "overthelex/multi-legal-bench", "hfDatasetDownloads": 224, "hfDatasetLikes": 0, "hfPaperUpvotes": 0, "hfPaperUrl": "https://huggingface.co/papers/2605.29738", "observedAt": "2026-10-04T03:02:42.754940Z", "signalStatus": {"githubStars": {"state": "not_applicable"}, "hfDatasetDownloads": {"checkedAt": "2026-10-04T15:31:48.030916Z", "lastSuccessfulDate": "2026-09-30", "state": "stale"}, "hfDatasetLikes": {"checkedAt": "2026-10-04T15:31:48.030916Z", "lastSuccessfulDate": "2026-09-30", "state": "stale"}, "hfPaperUpvotes": {"checkedAt": "2026-10-04T15:31:48.030916Z", "lastSuccessfulDate": "2026-09-30", "state": "stale"}}, "status": "stale"}, "capabilities": ["Reasoning"], "confidence": "High", "construction": "Aggregate Existing", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-08-20T15:22:10.044731Z", "inputHash": "932986e3935f9b1ed96a88e924160e52fb91ce946cbdbcbc32308527a3625f84", "model": "deepseek-v4-flash"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "Multi-Legal-Bench evaluates LLMs on legal reasoning across six countries, four language families, and five tasks, using structured metadata from court registries for classification and extraction.", "domainCuration": {"method": "rules-v1", "state": "auto"}, "evidence": {"reasonCodes": ["coined title prefix ending in Bench or Benchmark", "benchmark release stated in one sentence", "evaluation protocol evidence", "public artifact URL"], "snippet": "We introduce Multi-Legal-Bench, the first cross-jurisdictional legal benchmark that evaluates identical tasks across six countries (Ukraine, France, Netherlands, Poland, Czech Republic, Lithuania), four language families, and 165 million full-text court decisions."}, "familyId": "bmf_608eaa134d4c", "firstSeenAt": "2026-08-19", "githubIndex": {"candidateCount": 0, "indexedAt": "2026-09-17T04:40:34.111665+00:00", "repositories": [], "status": "not-found"}, "heat": null, "id": "bm_multi-legal-bench_2cfb13a9", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": ["Software & Cloud"], "links": {"code": null, "data": "https://huggingface.co/datasets/overthelex/multi-legal-bench", "hfPaper": "https://huggingface.co/papers/2605.29738", "paper": "https://arxiv.org/abs/2605.29738", "pdf": "https://arxiv.org/pdf/2605.29738", "project": null, "report": "https://arxiv.org/abs/2605.29738"}, "metrics": [], "motivation": "Legal NLP benchmarks overwhelmingly evaluate a single language or aggregate tasks that differ fundamentally across jurisdictions, making cross-lingual comparison impossible.", "name": "Multi-Legal-Bench", "oneLine": "Multi-Legal-Bench evaluates LLMs on legal reasoning across six countries, four language families, and five tasks, using structured metadata from court registries for classification and extraction.", "paperTitle": "Multi-Legal-Bench: Evaluating LLMs on Legal Reasoning Across Jurisdictions, Languages, and Legal Traditions", "primaryDomain": "Software & AI Compute", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2605.29738", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-10-14T01:46:58.058199Z"}, "publications": [], "ranking": {}, "readiness": "Inspectable", "recognitionConfidence": 1.0, "relation": "introduces", "releasedAt": "2026-05-28", "source": {"authors": ["Volodymyr Ovcharov"], "categories": ["cs.CL", "cs.AI"], "id": "2605.29738", "title": "Multi-Legal-Bench: Evaluating LLMs on Legal Reasoning Across Jurisdictions, Languages, and Legal Traditions", "type": "arxiv", "url": "https://arxiv.org/abs/2605.29738"}, "sourceUpdatedAt": "2026-08-07T16:55:49Z", "topics": ["Reasoning"], "venueAttempts": [], "whyItMatters": "It enables cross-lingual and cross-jurisdictional comparison of legal reasoning, revealing that transfer quality depends more on label-set alignment than language proximity, aiding model selection in legal NLP."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
