dp-schema.json

{
  "$id": "https://example.com/arrays.schema.json",
  "$schema": "http://json-schema.org/draft-07/schema#",
  "additionalProperties": false,
  "description": "Differentially Private Release",
  "properties": {
    "self": {
      "type": "object",
      "properties": {
        "schema_version": {
          "type": "string",
          "format": "uri"
        },
        "created": {
          "type": "string",
          "format": "date-time"
        },
        "description": {
          "type": "string"
        }
      },
      "required": [
        "schema_version",
        "created"
      ]
    },
    "dataset": {
      "additionalProperties": false,
      "title": "Dataset Level Metadata",
      "type": "object",
      "properties": {
        "description": {
          "type": "string"
        },
        "unitOfAnalysis": {
          "type": "string"
        },
        "structure": {
          "type": "string",
          "enum": [
            "wide",
            "long",
            "unknown"
          ]
        },
        "rowCount": {
          "description": "Number of rows or observations in the dataset",
          "oneOf": [
            {
              "type": "integer"
            },
            {
              "type": "string",
              "enum": [
                "private"
              ]
            }
          ]
        },
        "variableCount": {
          "type": "integer",
          "description": "Number of variables or features or columns in the dataset"
        },
        "dataSource": {
          "description": "Object containing descriptions of original source of dataset",
          "type": [
            "object",
            "null"
          ],
          "oneOf": [
            {
              "properties": {
                "name": {
                  "description": "Name such as a file name or database table name",
                  "type": "string"
                },
                "type": {
                  "description": "Type of data source",
                  "type": "string",
                  "enum": [
                    "File",
                    "SQL database"
                  ]
                },
                "format": {
                  "description": "Pertains to the type.  For example, a file that is \"text/csv\" or a SQL database that is \"Postgres version 9.1\"",
                  "type": "string"
                },
                "filesize": {
                  "description": "Filesize in bytes. (optional)",
                  "type": "integer"
                }
              },
              "required": [
                "name",
                "type"
              ]
            },
            {
              "type": "null"
            }
          ]
        },
        "citation": {
          "description": "Object containing a schema.org Dataset description",
          "type": "object",
          "properties": {
            "name": {
              "description": "Name of the dataset",
              "type": "string"
            }
          }
        }
      },
      "required": [
        "description",
        "unitOfAnalysis",
        "structure",
        "rowCount",
        "variableCount"
      ]
    },
    "dp-releases": {
      "items": {
        "$ref": "#/definitions/release"
      },
      "type": "array"
    }
  },
  "type": "object",
  "definitions": {
    "pure": {
      "type": "object",
      "required": [
        "name",
        "epsilon"
      ],
      "properties": {
        "name": {
          "type": "string",
          "enum": [
            "pure"
          ]
        },
        "epsilon": {
          "description": "Reflects the privacy cost of the query",
          "type": "number",
          "exclusiveMinimum": 0
        }
      },
      "additionalProperties": false
    },
    "approx": {
      "type": "object",
      "additionalProperties": false,
      "required": [
        "epsilon",
        "delta"
      ],
      "properties": {
        "name": {
          "type": "string",
          "enum": [
            "approx dp"
          ]
        },
        "epsilon": {
          "description": "Reflects the privacy cost of the query",
          "type": "number",
          "exclusiveMinimum": 0
        },
        "delta": {
          "description": "Reflects the privacy cost of the query",
          "maximum": 1,
          "minimum": 0,
          "type": "number"
        }
      }
    },
    "concentrated": {
      "type": "object",
      "required": [
        "name",
        "rho"
      ],
      "additionalProperties": false,
      "properties": {
        "name": {
          "type": "string",
          "enum": [
            "concentrated dp"
          ]
        },
        "rho": {
          "description": "dgjg",
          "exclusiveMinimum": 0,
          "type": "number"
        }
      }
    },
    "releaseInfo": {
      "type": "object",
      "additionalProperties": false,
      "properties": {
        "mechanism": {
          "enum": [
            "Laplace",
            "Gaussian",
            "Bootstrap"
          ],
          "type": "string"
        },
        "releaseValue": {}
      }
    },
    "release": {
      "type": "object",
      "properties": {
        "batch": {
          "type": "integer"
        },
        "nodeId": {
          "type": "integer",
          "minimum": 0,
          "description": "Determins which node the relase come from"
        },
        "accuracy": {
          "additionalProperties": false,
          "type": "object",
          "description": "accuracy information:",
          "properties": {
            "accuracyValue": {
              "type": "number"
            },
            "alpha": {
              "description": "The statistical significance level used to calculate accuracy",
              "type": "number"
            }
          }
        },
        "algorithmInfo": {
          "type": "object",
          "description": "Information about the algorithm",
          "additionalProperties": false,
          "properties": {
            "name": {
              "type": "string"
            },
            "cite": {
              "type": "string"
            },
            "argument": {
              "required": [
                "constraint",
                "n"
              ],
              "type": "object",
              "properties": {
                "constraint": {
                  "description": "User-provided clipping bounds for the data",
                  "properties": {
                    "lowerbound": {
                      "description": "Least numeric value that must be given by a user",
                      "type": "number"
                    },
                    "upperbound": {
                      "description": "Largest numeric value that must be given by a user",
                      "type": "number"
                    }
                  },
                  "required": [
                    "lowerbound",
                    "upperbound"
                  ],
                  "type": "object",
                  "value": [
                    "number",
                    "string"
                  ]
                },
                "n": {
                  "description": "The number of observations in the data",
                  "type": "number"
                }
              },
              "additionalProperties": true
            }
          },
          "required": [
            "argument",
            "name"
          ]
        },
        "privacyLoss": {
          "type": "object",
          "description": "The differential privacy definition",
          "oneOf": [
            {
              "$ref": "#/definitions/pure"
            },
            {
              "$ref": "#/definitions/approx"
            },
            {
              "$ref": "#/definitions/concentrated"
            }
          ]
        },
        "releaseInfo": {
          "$ref": "#/definitions/releaseInfo"
        },
        "statistic": {
          "enum": [
            "mean",
            "histogram",
            "covariance",
            "variance"
          ],
          "type": "string"
        },
        "variables": {
          "description": "This describes the column/s in a dataset..",
          "items": {
            "type": "string"
          },
          "type": "array"
        }
      },
      "required": [
        "statistic",
        "variables",
        "releaseInfo",
        "postprocess",
        "algorithmInfo",
        "accuracy",
        "batch",
        "privacyLoss"
      ]
    }
  },
  "required": [
    "self",
    "dataset",
    "dp-releases"
  ]
}