Untag a Dataset (v2)

DELETE /datasets/{dataset_id}/tags?tag=tag_string

Read notes about this design choice at the bottom.

Authentication

This endpoint requires an API key. Include it in a X-API-Key header on each request.

Name	Type	Required	Description
`dataset_id`	integer	Yes	The unique ID of the dataset to add a tag to

Name	Type	Required	Description
`tag`	string	Yes	tag to be removed for this dataset

Name	Required	Description
`X-API-Key`	Yes	your API Key
`Accept`	No	`application/json` recommended

None.

curl -X DELETE https://www.openml.org/api/v2/json/datasets/1234/tags?tag=biology \
 -H 'X-API-Key: YOUR_API_KEY' \
 -H 'Accept: application/json'

204 No Content - tag removed successfully (or tag was not present - since this endpoint is idempotent)

No response body.

Status	Code	Meaning
`400 Bad Request`	`VALIDATION_ERROR`	Could be a missing `tag` in query parameter, etc (will be specified in the error message)
`401 Unauthorized`	`UNAUTHENTICATED`	The Authorization header is missing or the API key is invalid.
`403 Forbidden`	`FORBIDDEN`	Dataset belongs to another user
`404 Not Found`	`DATASET_NOT_FOUND`	No dataset found with id dataset_id
`429 Too Many Requests`	`RATE_LIMITED`	You've exceeded our request rate limit. Check the `Retry-After` header before retrying.
`500 Internal Server Error`	`INTERNAL_ERROR`	Something went wrong on our end. You may retry your request.

Error responses with the following form:

{
  "error": {
    "code": "DATASET_NOT_FOUND",
    "message": "No dataset found with id 00000"
  }
}

Example - missing tag in request:

{
  "error": {
    "code": "VALIDATION_ERROR",
    "message": "No tag was specified."
  }
}

curl -X 'POST' \
  'https://www.openml.org/api/v1/json/data/untag?data_id=1234&tag=biology&api_key=12345' \
  -H 'accept: application/json' \
  -d ''

{
  "data_untag": {
    "id":"1234",
    "tag": ["health", "medical"]
  },
}

Notable changes:

General Notes:

Choosing DELETE /datasets/:dataset_id/tags?tag=tag_string versus DELETE /datasets/:dataset_id/tags/:tag where :tag is the tag_string, e.g. "biology", because we already have many tags that may have various formats. The format may not work easily as a path parameter and could require URL encoding from the client. That seems cumbersome for now, and other approaches also add more development burden, so I chose the option of including the string as a query parameter. All other points match general RESTful design patterns.