Retrieve Video Classification for DSP Pre-Bid Targeting

Verity provides brand safety and contextual classification data for partner video platforms such as IRIS.TV and JW Player.

For the purposes of pre-bid targeting, DSPs can get classification data for video assets on these platforms directly from Verity. A DSP can use the unique identifier assigned by the partner platform (such as the IRIS ID) to identify the video asset, without knowing the video URL.

Verity refers to the unique partner platform identifier for a video as the partner_id.

The integration works as follows:

• The partner video platform (such as IRIS.TV):

  • Initiates a video classification request to Verity, specifying the the video’s URL and partner_id (in this case the IRIS ID).

  • Verity maps the video URL and partner_id to a Verity uuid and analyzes the video.

  • Verity responds to IRIS.TV with the video URL, partner_id, uuid, and classification.

• A DSP partner (such as MediaMath):

  • Initiates a classification request to Verity, specifying a video partner_id (in this case the IRIS ID).

  • Verity checks for the video partner_id.

  • If data is available, Verity returns the video classification to the DSP.

Important: Contact verity-support@gumgum.com to find out more about enabling usage of partner_id for video classification.

Video Classification Endpoint for partner_id

The Verity video analysis retrieval endpoint is:

Example Video Request

This example walks through retrieving classification data for a video available in the IRIS.TV Contextual Marketplace. The video’s partner_id is an IRIS_ID of iris_12345.

Example GET Request

Submit a GET request to /video/classification/{partner_id}, for example:

curl -X GET "https://verity-api.gumgum.com/video/classification/iris_12345" -H "accept: /" -H "x-api-key: <YOUR_API_KEY_HERE>"

Note: The API Key you specify must include support for this feature.

Example JSON Response

If video classification data is available, the Verity API returns a JSON response with Content-Type: application/json; charset=UTF-8.

The data payload includes the JSON classification results detailing the complete brand safety, keyword, and categorization analysis data for the video.  For example:

{
  "uuid": "668e9d0e-a909-5ea0-9017-80c983a8db2c",
  "url": "https://vod-progressive.net/exp=1123~acl=%2Fvimeo-prod-skyfire-std-us%2F01%.mp4~hmac=a621155b76838fc3b5073dfdda3fbf730af5fd96d24bcbab3c25ee41dc023239/vimeo-prod-skyfire-std-us/01/3987/18/469935935/2090153996.mp4?download=1&filename=pexels-karolina-grabowska-5644289.mp4",
  "data": {
    "processedAt": "2021-05-21T16:47:35.352Z",
    "expiresAt": "2022-05-16T16:47:35.352Z",
    "languageCode": "en",
    "iab": {
      "v2": [
        {
          "id": "52",
          "category": "Business and Finance",
          "score": 0.95
        }
      ]
    },
    "keywords": [
      "black friday sale"
    ],
    "safe": true,
    "threats": [],
    "sentiments": [
      {
        "sentiment": "neutral",
        "score": 1
      },
      {
        "sentiment": "negative",
        "score": 0
      },
      {
        "sentiment": "positive",
        "score": 0
      }
    ]
  }
}

See JSON Response for details about the classification data included in the response.

More Information

• /wiki/spaces/VDC/pages/2336162016