{
  "$schema": "https://json-schema.org/draft/2020-12/schema",
  "$id": "https://acceldata.io/schemas/dataset.json",
  "title": "Dataset",
  "type": "object",
  "description": "A registered dataset in Acceldata",
  "properties": {
    "id": {
      "type": "string",
      "description": "Unique dataset identifier"
    },
    "name": {
      "type": "string",
      "description": "Dataset name (table or file name)"
    },
    "source": {
      "type": "string",
      "description": "Data source platform",
      "enum": [
        "snowflake",
        "databricks",
        "bigquery",
        "redshift",
        "s3",
        "hdfs",
        "hive",
        "postgres"
      ]
    },
    "database": {
      "type": "string",
      "description": "Database name"
    },
    "schema": {
      "type": "string",
      "description": "Schema name"
    },
    "qualityScore": {
      "type": "number",
      "format": "double",
      "description": "Overall data quality score (0-100)"
    },
    "lastScanned": {
      "type": "string",
      "format": "date-time",
      "description": "Last time the dataset was scanned"
    },
    "rowCount": {
      "type": "integer",
      "description": "Number of rows in the dataset"
    },
    "columnCount": {
      "type": "integer",
      "description": "Number of columns in the dataset"
    }
  }
}