{
  "schema_version": "2.0",
  "record_type": "dataset",
  "tier": "verified",
  "slug": "esl-bench",
  "name": "ESL-bench (Event-driven Synthetic Longitudinal Benchmark)",
  "summary": {
    "en": "License: conflict (NOASSERTION); Mdexa hosting: pending. the publisher's own statements contradict each other (permissive license vs restriction)",
    "zh": "许可证声明：冲突（NOASSERTION）；本站处理：待确认，暂只链接原出处。发布方自己的说法互相矛盾（宽松许可与限制并存）。"
  },
  "systems": [
    "modern-medicine"
  ],
  "domains": [
    "clinical-medicine"
  ],
  "content_types": [
    "benchmark-exam",
    "knowledge-graph"
  ],
  "content_description": {
    "en": "Virtual health user dataset for evaluating AI health assistants; each virtual user has a health profile, event timeline, clinical exam data, and knowledge-graph-grounded evaluation queries.",
    "zh": "用于评估AI健康助手的虚拟健康用户数据集：每个虚拟用户包含健康档案、事件时间线、临床检查数据和基于知识图谱的评估查询。"
  },
  "languages": [
    "en",
    "zh"
  ],
  "modalities": [
    "text",
    "tabular"
  ],
  "size": {
    "raw": "320 questions over 20 virtual users",
    "measures": []
  },
  "publisher": "mirobody",
  "year": {
    "value": 2026,
    "raw": "2026-11-02"
  },
  "license": {
    "legacy_status": "pending",
    "spdx": "NOASSERTION",
    "commercial_use": "conflict",
    "derivatives": "conflict",
    "redistribution": "conflict",
    "rationale": {
      "en": "the publisher's own statements contradict each other (permissive license vs restriction)",
      "zh": "发布方自己的说法互相矛盾（宽松许可与限制并存）"
    },
    "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: apache-2.0",
        "spdx": "Apache-2.0",
        "url": "https://huggingface.co/datasets/mirobody/ESL-Bench",
        "verified_on": "2026-11-12",
        "snapshot_id": "s01",
        "observed": "page",
        "captured_by": "pipeline"
      },
      {
        "location": "readme",
        "location_detail": {
          "en": "Statement found in snapshot s01 (readme)",
          "zh": "在快照 s01（readme）中找到的声明"
        },
        "excerpt": "[Apache 2.0](https://opensource.org/licenses/Apache-2.0)",
        "spdx": "Apache-2.0",
        "url": "https://huggingface.co/datasets/mirobody/ESL-Bench",
        "verified_on": "2026-11-12",
        "snapshot_id": "s01",
        "observed": "page",
        "captured_by": "pipeline"
      },
      {
        "location": "readme",
        "location_detail": {
          "en": "Statement found in snapshot s01 (readme)",
          "zh": "在快照 s01（readme）中找到的声明"
        },
        "excerpt": "> ⚠️ **Research use only.** Outputs are synthetic and intended for benchmarking AI agents. They should **not** be used for diagnosis or treatment decisions.",
        "url": "https://huggingface.co/datasets/mirobody/ESL-Bench",
        "verified_on": "2026-11-12",
        "snapshot_id": "s01",
        "observed": "page",
        "captured_by": "pipeline"
      },
      {
        "location": "dataset-page",
        "location_detail": {
          "en": "Statement found in snapshot s02 (page)",
          "zh": "在快照 s02（page）中找到的声明"
        },
        "excerpt": "⚠️ Research use only. Outputs are synthetic and intended for benchmarking AI agents. They should not be used for diagnosis or treatment decisions.",
        "url": "https://huggingface.co/datasets/mirobody/ESL-Bench",
        "verified_on": "2026-11-12",
        "snapshot_id": "s02",
        "observed": "page",
        "captured_by": "pipeline"
      }
    ]
  },
  "license_status": "conflict",
  "hosting": "pending",
  "hosting_reason": {
    "en": "License conflict: the publisher's own statements contradict each other (permissive license vs restriction). Link only.",
    "zh": "许可证冲突：发布方自己的说法互相矛盾（宽松许可与限制并存）。只链接原出处。"
  },
  "signoff": null,
  "risk_flags": [],
  "provenance": {
    "source_description": null,
    "upstream": []
  },
  "known_issues": [
    {
      "en": "Ground-truth answers for the knowledge-graph evaluation queries are released only after the competition closes.",
      "zh": "知识图谱评估问题的标准答案仅在比赛结束后公布。"
    },
    {
      "en": "Scores are not directly comparable between successive batches because the question mix differs.",
      "zh": "由于题目构成不同，不同批次之间的分数不能直接比较。"
    }
  ],
  "usage_notes": [
    {
      "kind": "intended-purpose",
      "excerpt": "Outputs are synthetic and intended for benchmarking AI agents.",
      "url": "https://huggingface.co/datasets/mirobody/ESL-Bench",
      "snapshot_id": "s01"
    },
    {
      "kind": "disclaimer",
      "excerpt": "They should not be used for diagnosis or treatment decisions.",
      "url": "https://huggingface.co/datasets/mirobody/ESL-Bench",
      "snapshot_id": "s02"
    },
    {
      "kind": "intended-purpose",
      "excerpt": "Outputs are synthetic and intended for benchmarking AI agents.",
      "url": "https://huggingface.co/datasets/mirobody/ESL-Bench",
      "snapshot_id": "s01"
    },
    {
      "kind": "disclaimer",
      "excerpt": "They should not be used for diagnosis or treatment decisions.",
      "url": "https://huggingface.co/datasets/mirobody/ESL-Bench",
      "snapshot_id": "s02"
    }
  ],
  "preflight": {
    "status": "not-run"
  },
  "access": {
    "mdexa_download": null,
    "external_urls": [
      "https://huggingface.co/datasets/mirobody/ESL-Bench"
    ],
    "notes": null
  },
  "citation": {
    "text": null,
    "references": [
      "arXiv:2604.02834"
    ]
  },
  "verification_level": "verified",
  "last_verified": "2026-11-14",
  "tags": [
    {
      "facet": "task_format",
      "value": "open-qa",
      "basis": "inferred",
      "inferred_from": "card-content-description",
      "excerpt": "Produces a **knowledge-graph-grounded objective evaluation query set** with deterministic ground truth",
      "snapshot_id": "s01"
    },
    {
      "facet": "task_format",
      "value": "multi-turn-dialogue",
      "basis": "inferred",
      "inferred_from": "card-content-description",
      "excerpt": "plant a premise in the conversation and check whether the agent resists fabricated premises without becoming over-skeptical of true ones",
      "snapshot_id": "s01"
    },
    {
      "facet": "intended_use",
      "value": "evaluation",
      "basis": "author-stated",
      "excerpt": "is a virtual health user dataset for evaluating AI health assistants",
      "snapshot_id": "s01"
    },
    {
      "facet": "specialty",
      "value": "multi-specialty",
      "basis": "inferred",
      "inferred_from": "card-content-description",
      "excerpt": "Each virtual user contains a complete health profile, event timeline, clinical exam data, and knowledge-graph-grounded evaluation queries",
      "snapshot_id": "s01"
    },
    {
      "facet": "source_type",
      "value": "knowledge-base",
      "basis": "inferred",
      "inferred_from": "card-content-description",
      "excerpt": "with deterministic ground truth derived from the event-indicator knowledge graph",
      "snapshot_id": "s01"
    },
    {
      "facet": "annotation_depth",
      "value": "with-explanation",
      "basis": "inferred",
      "inferred_from": "card-content-description",
      "excerpt": "Records a structured **event log** explaining why indicators change",
      "snapshot_id": "s01"
    },
    {
      "facet": "data_format",
      "value": "jsonl",
      "basis": "inferred",
      "inferred_from": "snapshot-structure",
      "excerpt": "path: data/202610/sample320-20261030.jsonl",
      "snapshot_id": "s01"
    },
    {
      "facet": "data_format",
      "value": "json",
      "basis": "inferred",
      "inferred_from": "snapshot-structure",
      "excerpt": "├── profile.json               # User profile (demographics, history, family history)",
      "snapshot_id": "s01"
    },
    {
      "facet": "official_split",
      "value": "eval-only",
      "basis": "inferred",
      "inferred_from": "snapshot-structure",
      "excerpt": "split: test",
      "snapshot_id": "s01"
    },
    {
      "facet": "benchmark_status",
      "value": "public-leaderboard",
      "basis": "author-stated",
      "excerpt": "**2026-11-02 — New leaderboard dataset: `sample320-20261030`**",
      "snapshot_id": "s01"
    }
  ],
  "changelog": [
    {
      "date": "2026-11-13",
      "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）。"
      }
    },
    {
      "date": "2026-11-13",
      "change": {
        "en": "Re-adjudicated under the current rules; all machine checks still pass.",
        "zh": "按当前规则重新判定；机器检查仍全部通过。"
      }
    },
    {
      "date": "2026-11-13",
      "change": {
        "en": "Re-adjudicated under the current rules; all machine checks still pass.",
        "zh": "按当前规则重新判定；机器检查仍全部通过。"
      }
    },
    {
      "date": "2026-11-13",
      "change": {
        "en": "Re-adjudicated under the current rules; all machine checks still pass.",
        "zh": "按当前规则重新判定；机器检查仍全部通过。"
      }
    },
    {
      "date": "2026-11-13",
      "change": {
        "en": "Re-adjudicated under the current rules; all machine checks still pass.",
        "zh": "按当前规则重新判定；机器检查仍全部通过。"
      }
    },
    {
      "date": "2026-11-13",
      "change": {
        "en": "Re-adjudicated under the current rules; all machine checks still pass.",
        "zh": "按当前规则重新判定；机器检查仍全部通过。"
      }
    },
    {
      "date": "2026-11-13",
      "change": {
        "en": "Re-adjudicated under the current rules; all machine checks still pass.",
        "zh": "按当前规则重新判定；机器检查仍全部通过。"
      }
    },
    {
      "date": "2026-11-13",
      "change": {
        "en": "Re-adjudicated under the current rules; all machine checks still pass.",
        "zh": "按当前规则重新判定；机器检查仍全部通过。"
      }
    },
    {
      "date": "2026-11-13",
      "change": {
        "en": "Re-adjudicated under the current rules; all machine checks still pass.",
        "zh": "按当前规则重新判定；机器检查仍全部通过。"
      }
    },
    {
      "date": "2026-11-13",
      "change": {
        "en": "Re-adjudicated under the current rules; all machine checks still pass.",
        "zh": "按当前规则重新判定；机器检查仍全部通过。"
      }
    },
    {
      "date": "2026-11-13",
      "change": {
        "en": "Re-adjudicated under the current rules; all machine checks still pass.",
        "zh": "按当前规则重新判定；机器检查仍全部通过。"
      }
    },
    {
      "date": "2026-11-14",
      "change": {
        "en": "Re-adjudicated under the current rules; all machine checks still pass.",
        "zh": "按当前规则重新判定；机器检查仍全部通过。"
      }
    },
    {
      "date": "2026-11-14",
      "change": {
        "en": "Re-adjudicated under the current rules; all machine checks still pass.",
        "zh": "按当前规则重新判定；机器检查仍全部通过。"
      }
    }
  ],
  "access_recipe": {
    "verified_on": "2026-11-14",
    "platform": "huggingface",
    "repo_id": "mirobody/ESL-Bench",
    "source": {
      "api_url": "https://huggingface.co/api/datasets/mirobody/ESL-Bench",
      "fetched_on": "2026-11-14"
    },
    "access_state": "open",
    "files": [],
    "files_truncated": false,
    "commands": [
      {
        "command": "hf download mirobody/ESL-Bench --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": "efde471bc9f725f79bb12c7e123409bf0e662841",
      "last_modified": "2026-11-02T06:44:31.000Z"
    },
    "license_note": {
      "license_status": "conflict",
      "text": "Use according to the terms of the official source; this recipe does not say the data can be used freely."
    },
    "gaps": [
      "the platform publishes no checksum for these files; none was computed (we do not download data files)"
    ]
  },
  "_links": {
    "text": "Continue in this database. This record is one of 109 processed and verified records (datasets, data resources and open models). To find others, call `https://mcp.mdexa.com/v1/datasets?q=<words>` or connect the MCP server at `https://mcp.mdexa.com/mcp`; no key is needed for 20 calls per minute. Rows are under 1 KB; an empty result means \"not listed\". For a list of items, one batch check (free account) returns a report. Guide: /agents.md. Similar records: CMB (Chinese Medical Benchmark) (https://mdexa.com/datasets/cmb/); MedMCQA (https://mdexa.com/datasets/medmcqa/); PromptCBLUE (https://mdexa.com/datasets/promptcblue/).",
    "records": 109,
    "search": {
      "template": "https://mcp.mdexa.com/v1/datasets?q={words}",
      "example": "https://mcp.mdexa.com/v1/datasets?q=binding%20affinity"
    },
    "mcp": "https://mcp.mdexa.com/mcp",
    "anonymous_calls_per_minute": 20,
    "guide": "https://mdexa.com/agents.md",
    "faq": "https://mdexa.com/developers/faq/",
    "similar": [
      {
        "name": "CMB (Chinese Medical Benchmark)",
        "url": "https://mdexa.com/datasets/cmb/",
        "json": "https://mdexa.com/datasets/cmb.json"
      },
      {
        "name": "MedMCQA",
        "url": "https://mdexa.com/datasets/medmcqa/",
        "json": "https://mdexa.com/datasets/medmcqa.json"
      },
      {
        "name": "PromptCBLUE",
        "url": "https://mdexa.com/datasets/promptcblue/",
        "json": "https://mdexa.com/datasets/promptcblue.json"
      }
    ]
  }
}