Brand safety engines

Brand safety engines process media to determine where content falls on a scale of sensitivity or concern. This scale is composed of 11 designated sensitive topics created by the Global Alliance for Responsible Media (GARM), a social media and advertising industry body that helps standardize measures for insulating brands from harmful content posted to their sites and apps.

Engines used

The following engines are used to build a brand safety job:

Brand safety job

Enter the GraphQL mutation below. Update the payload data with your values.

This will create a job that will take a payload consisting of the media's URL, and optionally an array of webhooks for notification and additional processing.

GraphQL mutation:

mutation brandSafety(
  $url: String!,
  $notificationUris: [String]
) {
  createJob(
    input: {
      target: { name: "BrandSafety" }
      clusterId: "prd5-21xbyq0x-4h0s-o685-snas-oovhdai552v9"
      tasks: [
        {
          # Stream Ingestor  
          engineId: "8bdb0e3b-ff28-4f6e-a3ba-887bd06e6440"
          payload: { url: $url, ffmpegTemplate: "rawchunk" }
          ioFolders: [{ referenceId: "chunkOutput", mode: chunk, type: output }]
          
        }
        {
          # Transcription E7: English (Global) 
          engineId: "c0e55cde-340b-44d7-bb42-2e0d65e98255" 
          ioFolders: [
            { referenceId: "transcriptionInput", mode: chunk, type: input }
            { referenceId: "transcriptionOutput", mode: chunk, type: output }
          ]
          executionPreferences: {
            maxEngines: 10
            parentCompleteBeforeStarting: true
          }
        }
        {
          engineId: "8eccf9cc-6b6d-4d7d-8cb3-7ebf4950c5f3" # outputwriter
          ioFolders: [
            { referenceId: "owTranscriptionInput", mode: chunk, type: input }
          ]
          executionPreferences: {
            parentCompleteBeforeStarting: true
          }
        }
        {
          #Chunk Aggregator, used to aggregate words from AION - Standard transcript into text
          engineId: "045e675b-ed7c-43bf-8300-a2f2102f57a4" 
          payload: { engineOutputOption: "text" }
          executionPreferences: {
            parentCompleteBeforeStarting: true
          }
          ioFolders: [
            { referenceId: "textCreatorInput", mode: chunk, type: input }
            { referenceId: "textCreatorOutput", mode: chunk, type: output }
          ]
        }
        {
          # Entity Extraction G
          engineId: "cb2b77b9-3478-4e51-844f-e4064384bb33" 
          executionPreferences: {
            parentCompleteBeforeStarting: true
          }
          ioFolders: [
            { referenceId: "entityInput", mode: chunk, type: input }
            { referenceId: "entityOutput", mode: chunk, type: output }
          ]
        }
        {
          # Output writer
          engineId: "8eccf9cc-6b6d-4d7d-8cb3-7ebf4950c5f3"
          ioFolders: [
            { referenceId: "owEntityInput", mode: chunk, type: input }
          ]
          executionPreferences: {
            parentCompleteBeforeStarting: true
          }
        }
        {
          # Output writer
          engineId: "8eccf9cc-6b6d-4d7d-8cb3-7ebf4950c5f3"
          ioFolders: [
            { referenceId: "owClassificationInput", mode: chunk, type: input }
          ]
          executionPreferences: {
            parentCompleteBeforeStarting: true
          }
        }
        {
          # Classification AC: IAB
          engineId: "06bbc2e7-aa59-4c32-9ec3-1a147fff78a6" 
          executionPreferences: {
            parentCompleteBeforeStarting: true
          }
          ioFolders: [
            { referenceId: "classificationInput", mode: chunk, type: input }
            { referenceId: "classificationOutput", mode: chunk, type: output }
          ]
        }
      ]
      notificationUris: $notificationUris
      routes: [
        {
          parentIoFolderReferenceId: "chunkOutput"
          childIoFolderReferenceId: "transcriptionInput"
        }
        {
          parentIoFolderReferenceId: "transcriptionOutput"
          childIoFolderReferenceId: "textCreatorInput"
        }
        {
          parentIoFolderReferenceId: "textCreatorOutput"
          childIoFolderReferenceId: "entityInput"
        }
        {
          parentIoFolderReferenceId: "transcriptionOutput"
          childIoFolderReferenceId: "owTranscriptionInput"
        }
        {
          parentIoFolderReferenceId: "textCreatorOutput"
          childIoFolderReferenceId: "classificationInput"
        }
        {
          parentIoFolderReferenceId: "entityOutput"
          childIoFolderReferenceId: "owEntityInput"
        }
        {
          parentIoFolderReferenceId: "classificationOutput"
          childIoFolderReferenceId: "owClassificationInput"
        }
      ]
    }
  ) {
    id
    targetId
  }
}

GraphQL example variables for the mutation:

{
  "url": "https://example.com/path-to-media",
  "notificationUris": ["https://webhook.example.com"]
}

GraphQL response:

{
  "data": {
    "createJob": {
      "id": "22030903_Gq1GRNdmwc",
      "targetId": "1951002277",
     }
  }
}

Note the job ID (id) and target ID (targetId) for progress and result checks.

Next steps

Now, query the job status and results.

• Using GraphQL
• Global Alliance for Responsible Media (GARM)
• See the cognitive engine classes
• See an overview of cognitive technology
• View engines for your media type

• Using aiWARE
• Use cognitive engines
• Choose an engine
• Capability
• Data
• Brand safety

• Engines used
• Brand safety job
• Next steps