{
  "schema_version": "2.0",
  "record_type": "dataset",
  "tier": "verified",
  "slug": "pubmedcausal",
  "name": "PubMedCausal",
  "summary": {
    "en": "License: permissive (MIT); Mdexa hosting: link-only. first-hand data-scope statement(s) allow commercial use and redistribution",
    "zh": "许可证声明：宽松（MIT）；本站处理：只链接原出处。针对数据的声明允许商用和再分发。"
  },
  "systems": [
    "modern-medicine"
  ],
  "domains": [
    "biomedical-literature"
  ],
  "content_types": [
    "annotated-corpus",
    "literature"
  ],
  "content_description": {
    "en": "Span-level annotated corpus for causal relation extraction in biomedical text, built from 30,000 paragraph-level PubMed abstract samples with 6,491 adjudicated cause-effect pairs.",
    "zh": "面向生物医学文本因果关系抽取的跨度级标注语料，含3万段PubMed摘要样本与6491个判定因果关系对。"
  },
  "languages": [
    "en"
  ],
  "modalities": [
    "text"
  ],
  "size": {
    "raw": "30,000 paragraph-level samples",
    "measures": []
  },
  "publisher": "jaypee01",
  "year": {
    "value": null,
    "raw": null
  },
  "license": {
    "legacy_status": "pending",
    "spdx": "MIT",
    "commercial_use": "yes",
    "derivatives": "yes",
    "redistribution": "yes",
    "rationale": {
      "en": "first-hand data-scope statement(s) allow commercial use and redistribution",
      "zh": "针对数据的声明允许商用和再分发"
    },
    "evidence": [
      {
        "location": "hf-card-metadata",
        "location_detail": {
          "en": "Statement found in snapshot s01 (readme)",
          "zh": "在快照 s01（readme）中找到的声明"
        },
        "excerpt": "license: mit",
        "spdx": "MIT",
        "url": "https://huggingface.co/datasets/jaypee01/PubMedCausal",
        "verified_on": "2026-11-12",
        "snapshot_id": "s01",
        "observed": "page",
        "captured_by": "pipeline"
      }
    ],
    "license_conditions": [
      "notice-required"
    ]
  },
  "license_status": "permissive",
  "hosting": "link-only",
  "hosting_reason": {
    "en": "License permissive: first-hand data-scope statement(s) allow commercial use and redistribution. Link only.",
    "zh": "许可证宽松：针对数据的声明允许商用和再分发。只链接原出处。"
  },
  "signoff": null,
  "risk_flags": [],
  "provenance": {
    "source_description": null,
    "upstream": []
  },
  "known_issues": [
    {
      "en": "Severe class imbalance: only 13.15% of samples contain annotated causality, so accuracy alone is misleading.",
      "zh": "类别严重不平衡：仅13.15%的样本含标注因果关系，单看准确率会失真。"
    },
    {
      "en": "Annotated relations express causal claims in the literature; they do not establish that the claims are true, and predictions are not medical advice.",
      "zh": "标注关系仅表示文献中的因果主张，并不证明其成立，模型预测不可作为医疗建议。"
    }
  ],
  "preflight": {
    "status": "not-run"
  },
  "access": {
    "mdexa_download": null,
    "external_urls": [
      "https://huggingface.co/datasets/jaypee01/PubMedCausal"
    ],
    "notes": null
  },
  "citation": {
    "text": null,
    "references": []
  },
  "verification_level": "verified",
  "last_verified": "2026-11-14",
  "tags": [
    {
      "facet": "task_format",
      "value": "relation-extraction",
      "basis": "author-stated",
      "excerpt": "**PubMedCausal** is a span-level annotated corpus for **causal relation extraction in biomedical text**.",
      "snapshot_id": "s01"
    },
    {
      "facet": "task_format",
      "value": "text-classification",
      "basis": "author-stated",
      "excerpt": "- text-classification",
      "snapshot_id": "s01"
    },
    {
      "facet": "intended_use",
      "value": "evaluation",
      "basis": "author-stated",
      "excerpt": "These properties make the corpus suitable for evaluating causal extraction under conditions that more closely resemble causal statements encountered in biomedical literature.",
      "snapshot_id": "s01"
    },
    {
      "facet": "intended_use",
      "value": "fine-tuning",
      "basis": "inferred",
      "inferred_from": "card-content-description",
      "excerpt": "Models trained exclusively on PubMedCausal may not generalize directly to causal language in domains such as finance, law, news, or everyday discourse.",
      "snapshot_id": "s01"
    },
    {
      "facet": "specialty",
      "value": "multi-specialty",
      "basis": "inferred",
      "inferred_from": "card-content-description",
      "excerpt": "PubMedCausal was constructed from biomedical research abstracts indexed in **PubMed**.",
      "snapshot_id": "s01"
    },
    {
      "facet": "source_type",
      "value": "literature",
      "basis": "author-stated",
      "excerpt": "The corpus is derived from PubMed abstracts.",
      "snapshot_id": "s01"
    },
    {
      "facet": "expert_involvement",
      "value": "review-unspecified",
      "basis": "author-stated",
      "excerpt": "Annotations were subsequently reviewed and adjudicated to produce the final corpus.",
      "snapshot_id": "s01"
    },
    {
      "facet": "official_split",
      "value": "full-split",
      "basis": "author-stated",
      "excerpt": "The Hugging Face release provides train and test partitions for the supported task configurations.",
      "snapshot_id": "s01"
    }
  ],
  "changelog": [
    {
      "date": "2026-11-14",
      "change": {
        "en": "Added via the automated pipeline and passed all checks (gates, verbatim quotes, rule engine, two independent extractions).",
        "zh": "经自动流水线收录并通过全部校验（门禁、逐字引用、判定引擎、两轮独立抽取）。"
      }
    }
  ],
  "access_recipe": {
    "verified_on": "2026-11-14",
    "platform": "huggingface",
    "repo_id": "jaypee01/PubMedCausal",
    "source": {
      "api_url": "https://huggingface.co/api/datasets/jaypee01/PubMedCausal",
      "fetched_on": "2026-11-14"
    },
    "access_state": "open",
    "files": [
      {
        "name": ".gitattributes",
        "url": "https://huggingface.co/datasets/jaypee01/PubMedCausal/resolve/fb01de12ea3d12fe45a5a5ba173c8d113a00e949/.gitattributes",
        "size_bytes": 2639,
        "checksum": null,
        "format": "gitattributes"
      },
      {
        "name": "LICENSE",
        "url": "https://huggingface.co/datasets/jaypee01/PubMedCausal/resolve/fb01de12ea3d12fe45a5a5ba173c8d113a00e949/LICENSE",
        "size_bytes": 1069,
        "checksum": null,
        "format": null
      },
      {
        "name": "PubMedCausaldata/30k_test.json",
        "url": "https://huggingface.co/datasets/jaypee01/PubMedCausal/resolve/fb01de12ea3d12fe45a5a5ba173c8d113a00e949/PubMedCausaldata/30k_test.json",
        "size_bytes": 28269360,
        "checksum": {
          "algorithm": "sha256",
          "value": "ee4db1cea406a7387ba6696310402c5c2eb31a2cc584e42419117446c6f9e68b"
        },
        "format": "json"
      },
      {
        "name": "PubMedCausaldata/30k_train.json",
        "url": "https://huggingface.co/datasets/jaypee01/PubMedCausal/resolve/fb01de12ea3d12fe45a5a5ba173c8d113a00e949/PubMedCausaldata/30k_train.json",
        "size_bytes": 28277345,
        "checksum": {
          "algorithm": "sha256",
          "value": "4cd866724d3624099f73237f1af1ec959a5a2744c607f76fc9862606459ef9b7"
        },
        "format": "json"
      },
      {
        "name": "PubMedCausaldata/validation_set.json",
        "url": "https://huggingface.co/datasets/jaypee01/PubMedCausal/resolve/fb01de12ea3d12fe45a5a5ba173c8d113a00e949/PubMedCausaldata/validation_set.json",
        "size_bytes": 2052431,
        "checksum": null,
        "format": "json"
      },
      {
        "name": "README.md",
        "url": "https://huggingface.co/datasets/jaypee01/PubMedCausal/resolve/fb01de12ea3d12fe45a5a5ba173c8d113a00e949/README.md",
        "size_bytes": 10801,
        "checksum": null,
        "format": "md"
      }
    ],
    "files_truncated": false,
    "commands": [
      {
        "command": "hf download jaypee01/PubMedCausal --repo-type dataset",
        "evidence": {
          "excerpt": "You can use the hf download command from the terminal to directly download files from the Hub.",
          "doc": "hf-download",
          "url": "https://huggingface.co/docs/huggingface_hub/guides/download",
          "fetched_on": "2026-11-14"
        }
      }
    ],
    "registration": null,
    "version": {
      "commit": "fb01de12ea3d12fe45a5a5ba173c8d113a00e949",
      "last_modified": "2026-10-06T13:02:30.000Z"
    },
    "license_note": {
      "license_status": "permissive",
      "text": "Use according to the terms of the official source; this recipe does not say the data can be used freely."
    },
    "gaps": []
  },
  "_links": {
    "text": "在本库继续查找。本条目是 109 条已处理并核实的记录之一（数据集、数据资源与开源模型）。查找其他条目：调用 `https://mcp.mdexa.com/v1/datasets?q=<关键词>`，或连接 MCP 服务 `https://mcp.mdexa.com/mcp`；每分钟 20 次以内无需密钥。每行返回不到 1 KB；空结果即\"数据库未收录\"。核对一份清单：一次批量核对（免费账号）即返回报告。说明：/agents.md。相似条目：CBLUE (https://mdexa.com/zh/datasets/cblue/)；PMC-Patients-ReCDS (https://mdexa.com/zh/datasets/pmc-patients-recds/)；PromptCBLUE (https://mdexa.com/zh/datasets/promptcblue/)。",
    "records": 109,
    "search": {
      "template": "https://mcp.mdexa.com/v1/datasets?q={words}",
      "example": "https://mcp.mdexa.com/v1/datasets?q=%E7%BB%93%E5%90%88%E4%BA%B2%E5%92%8C%E5%8A%9B"
    },
    "mcp": "https://mcp.mdexa.com/mcp",
    "anonymous_calls_per_minute": 20,
    "guide": "https://mdexa.com/zh/agents.md",
    "faq": "https://mdexa.com/zh/developers/faq/",
    "similar": [
      {
        "name": "CBLUE",
        "url": "https://mdexa.com/zh/datasets/cblue/",
        "json": "https://mdexa.com/zh/datasets/cblue.json"
      },
      {
        "name": "PMC-Patients-ReCDS",
        "url": "https://mdexa.com/zh/datasets/pmc-patients-recds/",
        "json": "https://mdexa.com/zh/datasets/pmc-patients-recds.json"
      },
      {
        "name": "PromptCBLUE",
        "url": "https://mdexa.com/zh/datasets/promptcblue/",
        "json": "https://mdexa.com/zh/datasets/promptcblue.json"
      }
    ]
  }
}