{
  "$schema": "https://json-schema.org/draft/2020-12/schema",
  "$id": "https://schema.govfresh.com/v1/catalog/dataset.schema.json",
  "title": "Dataset",
  "description": "One published dataset, aligned with DCAT and expressed as a schema.org Dataset. Every schemaGov profile a publisher serves should also appear here as a dataset, so that a consumer arriving through an open-data portal finds the structured data rather than only a CSV of it.",
  "standards": [
    "dcat"
  ],
  "type": "object",
  "properties": {
    "@context": {},
    "@id": {
      "type": "string",
      "format": "iri",
      "description": "This dataset's own dereferenceable identifier."
    },
    "@type": {
      "const": "Dataset",
      "description": "Always `Dataset`, a real schema.org type."
    },
    "name": {
      "type": "string",
      "description": "The dataset's title."
    },
    "description": {
      "type": "string",
      "description": "A plain-language description of what this dataset contains."
    },
    "url": {
      "type": "string",
      "format": "iri",
      "description": "Where this dataset's canonical human-readable page lives."
    },
    "identifier": {
      "type": "array",
      "description": "Any scheme-qualified identifiers for this dataset beyond its `@id`.",
      "items": {
        "$ref": "../_core/identifier.schema.json"
      }
    },
    "publisher": {
      "$ref": "../_core/reference.schema.json",
      "description": "The `GovernmentOrganization` publishing this dataset."
    },
    "creator": {
      "$ref": "../_core/reference.schema.json",
      "description": "Who produced the data, when different from the publisher."
    },
    "includedInDataCatalog": {
      "$ref": "../_core/reference.schema.json",
      "description": "The `Catalog` this dataset is listed in."
    },
    "keywords": {
      "type": "array",
      "description": "Search terms someone looking for this dataset might use.",
      "items": {
        "type": "string"
      }
    },
    "dataTheme": {
      "type": "array",
      "description": "termCode(s) from the DataTheme code list, aligned with the EU data-theme vocabulary that DCAT-AP uses.",
      "items": {
        "type": "string",
        "enum": [
          "AGRI",
          "ECON",
          "EDUC",
          "ENER",
          "ENVI",
          "GOVE",
          "HEAL",
          "INTR",
          "JUST",
          "OP_DATPRO",
          "REGI",
          "SOCI",
          "TECH",
          "TRAN"
        ]
      }
    },
    "updateFrequency": {
      "type": "string",
      "description": "How often this is refreshed, as an ISO 8601 duration (P1D, P1M) or 'continuous' or 'irregular'. DCAT dct:accrualPeriodicity. A dataset with no stated cadence cannot be told apart from an abandoned one."
    },
    "accessRights": {
      "type": "string",
      "enum": [
        "public",
        "restricted",
        "nonPublic"
      ],
      "description": "DCAT dct:accessRights. 'restricted' means the metadata is public but the data is not."
    },
    "spatialCoverage": {
      "$ref": "../_core/reference.schema.json",
      "description": "The `Jurisdiction` or area this dataset covers."
    },
    "temporalCoverage": {
      "type": "string",
      "description": "ISO 8601 interval, e.g. 2020-01-01/2025-12-31."
    },
    "datePublished": {
      "type": "string",
      "format": "date",
      "description": "When this dataset was first published."
    },
    "dateModified": {
      "type": "string",
      "anyOf": [
        {
          "format": "date"
        },
        {
          "format": "date-time"
        }
      ],
      "description": "When this dataset last changed."
    },
    "license": {
      "type": "string",
      "format": "iri",
      "description": "The licence this dataset is published under."
    },
    "inLanguage": {
      "type": "array",
      "description": "The language(s) this dataset is published in.",
      "items": {
        "type": "string"
      }
    },
    "conformsToProfile": {
      "type": "string",
      "description": "The schemaGov profile this dataset serves, where it serves one. This is what links an open-data catalogue entry to structured schemaGov data.",
      "enum": [
        "core",
        "org",
        "code",
        "meetings",
        "requests",
        "budget",
        "procurement",
        "catalog",
        "alerts"
      ]
    },
    "distribution": {
      "type": "array",
      "minItems": 1,
      "description": "The actual files or endpoints. A dataset with no distribution is an announcement, not data.",
      "items": {
        "type": "object",
        "properties": {
          "@type": {
            "const": "DataDownload"
          },
          "name": {
            "type": "string"
          },
          "contentUrl": {
            "type": "string",
            "format": "iri",
            "description": "DCAT dcat:downloadURL."
          },
          "accessUrl": {
            "type": "string",
            "format": "iri",
            "description": "DCAT dcat:accessURL: a landing page or API root, where there is no direct download."
          },
          "encodingFormat": {
            "type": "string"
          },
          "contentSize": {
            "type": "string"
          },
          "dateModified": {
            "type": "string",
            "anyOf": [
              {
                "format": "date"
              },
              {
                "format": "date-time"
              }
            ]
          }
        },
        "required": [
          "@type"
        ],
        "additionalProperties": false
      }
    },
    "conformsTo": {
      "type": "array",
      "items": {
        "type": "string"
      },
      "description": "External standards this dataset follows, e.g. an Akoma Ntoso or OpenAPI version. DCAT dct:conformsTo. Distinct from conformanceLevel."
    },
    "applicableLegislation": {
      "type": "array",
      "items": {
        "type": "string",
        "format": "iri"
      },
      "description": "Legislation mandating or governing this dataset, typically an ELI URI. Observed on EU portals under the Open Data Directive; a direct join from catalog to the code profile."
    },
    "hvdCategory": {
      "type": "string",
      "format": "iri",
      "description": "EU High Value Dataset category URI. A legal classification under the Open Data Directive, not a subject theme."
    },
    "highValueDataset": {
      "type": "boolean",
      "description": "Whether this dataset is designated a High Value Dataset under the EU Open Data Directive. See `hvdCategory` for which category."
    },
    "spatialReferenceSystem": {
      "type": "string",
      "description": "EPSG code of the dataset's coordinate reference system, e.g. 'epsg:2157'. GeoJSON mandates WGS84 but national grids are in daily use, so an international profile cannot assume it."
    },
    "provenanceStatement": {
      "type": "string",
      "description": "A plain-language note on where the data comes from and how it's produced, where that isn't obvious from the publisher alone."
    }
  },
  "required": [
    "@id",
    "@type",
    "name",
    "publisher",
    "distribution"
  ],
  "additionalProperties": true
}