@datafire/azure_cognitiveservices_personalizer

DataFire integration for Personalizer Client

Usage no npm install needed!

<script type="module">
  import datafireAzureCognitiveservicesPersonalizer from 'https://cdn.skypack.dev/@datafire/azure_cognitiveservices_personalizer';
</script>

README

@datafire/azure_cognitiveservices_personalizer

Client library for Personalizer Client

Installation and Usage

npm install --save @datafire/azure_cognitiveservices_personalizer

let azure_cognitiveservices_personalizer = require('@datafire/azure_cognitiveservices_personalizer').create({
  apim_key: ""
});

.then(data => {
  console.log(data);
});

Description

Personalizer Service is an Azure Cognitive Service that makes it easy to target content and experiences without complex pre-analysis or cleanup of past data. Given a context and featurized content, the Personalizer Service returns which content item to show to users in rewardActionId. As rewards are sent in response to the use of rewardActionId, the reinforcement learning algorithm will improve the model and improve performance of future rank calls.

Actions

Policy_Reset

Resets the Learning settings of the Personalizer service to default.

azure_cognitiveservices_personalizer.Policy_Reset(null, context)

Input

This action has no parameters

Output

Policy_Get

Get the Learning settings currently used by the Personalizer service.

azure_cognitiveservices_personalizer.Policy_Get(null, context)

Input

This action has no parameters

Output

Policy_Update

Update the Learning settings that the Personalizer service will use to train models.

azure_cognitiveservices_personalizer.Policy_Update({
  "policy": {
    "name": "",
    "arguments": ""
  }
}, context)

Input

Output

ServiceConfiguration_Get

Get the Personalizer service configuration.

azure_cognitiveservices_personalizer.ServiceConfiguration_Get(null, context)

Input

This action has no parameters

Output

ServiceConfiguration_Update

Update the Personalizer service configuration.

azure_cognitiveservices_personalizer.ServiceConfiguration_Update({
  "config": {
    "rewardWaitTime": "",
    "defaultReward": 0,
    "rewardAggregation": "",
    "explorationPercentage": 0,
    "modelExportFrequency": "",
    "logRetentionDays": 0
  }
}, context)

Input

Output

Evaluations_List

List all the submitted evaluations.

azure_cognitiveservices_personalizer.Evaluations_List(null, context)

Input

This action has no parameters

Output

Evaluations_Create

Submit a new evaluation job.

azure_cognitiveservices_personalizer.Evaluations_Create({
  "evaluation": {
    "name": "",
    "startTime": "",
    "endTime": "",
    "policies": []
  }
}, context)

Input

Output

Evaluations_Delete

Delete the evaluation associated with the Id.

azure_cognitiveservices_personalizer.Evaluations_Delete({
  "evaluationId": ""
}, context)

Input

  • input object
    • evaluationId required string: Id of the evaluation to delete.

Output

Output schema unknown

Evaluations_Get

Get the evaluation associated with the Id.

azure_cognitiveservices_personalizer.Evaluations_Get({
  "evaluationId": ""
}, context)

Input

  • input object
    • evaluationId required string: Id of the evaluation.

Output

Events_Activate

Report that the specified event was actually displayed to the user and a reward should be expected for it

azure_cognitiveservices_personalizer.Events_Activate({
  "eventId": ""
}, context)

Input

  • input object
    • eventId required string: The event ID this activation applies to.

Output

Output schema unknown

Events_Reward

Report reward that resulted from using the action specified in rewardActionId for the specified event.

azure_cognitiveservices_personalizer.Events_Reward({
  "eventId": "",
  "reward": {
    "value": 0
  }
}, context)

Input

  • input object
    • eventId required string: The event id this reward applies to.
    • reward required RewardRequest

Output

Output schema unknown

Log_Delete

Delete all generated logs.

azure_cognitiveservices_personalizer.Log_Delete(null, context)

Input

This action has no parameters

Output

Output schema unknown

Log_GetProperties

Get properties of generated logs.

azure_cognitiveservices_personalizer.Log_GetProperties(null, context)

Input

This action has no parameters

Output

Model_Reset

Resets the model file generated by Personalizer service.

azure_cognitiveservices_personalizer.Model_Reset(null, context)

Input

This action has no parameters

Output

Output schema unknown

Model_Get

Get the model file generated by Personalizer service.

azure_cognitiveservices_personalizer.Model_Get(null, context)

Input

This action has no parameters

Output

  • output file

Model_GetProperties

Get properties of the model file generated by Personalizer service.

azure_cognitiveservices_personalizer.Model_GetProperties(null, context)

Input

This action has no parameters

Output

Rank

Submit a Personalizer rank request, to get which of the provided actions should be used in the provided context.

azure_cognitiveservices_personalizer.Rank({
  "rankRequest": {
    "actions": []
  }
}, context)

Input

Output

Definitions

ContainerStatus

  • ContainerStatus object
    • apiStatus string
    • apiStatusMessage string
    • service string

DateRange

  • DateRange object
    • from string
    • to string

ErrorResponse

  • ErrorResponse object: Used to return an error to the client

Evaluation

  • Evaluation object
    • endTime string
    • featureImportance array
      • items array
        • items string
    • id string
    • jobId string
    • name string
    • policyResults array
    • startTime string
    • status string (values: completed, pending, failed, notSubmitted)

EvaluationContract

  • EvaluationContract object: A counterfactual evaluation.
    • enableOfflineExperimentation boolean: True if the evaluation should explore for a more optimal Learning settings.
    • endTime required string: The end time of the evaluation.
    • name required string: The name of the evaluation.
    • policies required array: Additional Learning settings to evaluate.
    • startTime required string: The start time of the evaluation.

InternalError

  • InternalError object: An object containing more specific information than the parent object about the error.

LogsProperties

  • LogsProperties object
    • dateRange
      • from string
      • to string

ModelProperties

  • ModelProperties object
    • creationTime string
    • lastModifiedTime string

PersonalizerError

  • PersonalizerError object: The error object.
    • code required string (values: BadRequest, ResourceNotFound, InternalServerError, InvalidServiceConfiguration, InvalidPolicyConfiguration, InvalidPolicyContract, InvalidEvaluationContract, InvalidRewardRequest, InvalidEventIdToActivate, InvalidRankRequest, InvalidExportLogsRequest, InvalidContainer, FrontEndNotFound, EvaluationNotFound, LogsPropertiesNotFound, RankNullResponse, UpdateConfigurationFailed, ModelResetFailed): High level error code.
    • details array: An array of details about specific errors that led to this reported error.
    • innerError InternalError
    • message required string: A message explaining the error reported by the service.
    • target string: Error source element.

PolicyContract

  • PolicyContract object: Learning settings specifying how to train the model.
    • arguments required string: Arguments of the Learning settings.
    • name required string: Name of the Learning settings.

PolicyResult

  • PolicyResult object
    • arguments string
    • name string
    • summary array
    • totalSummary
      • aggregateTimeWindow string
      • confidenceInterval number
      • ipsEstimatorDenominator number
      • ipsEstimatorNumerator number
      • nonZeroProbability number
      • snipsEstimatorDenominator number
      • sumOfSquares number
      • timeStamp string

PolicyResultSummary

  • PolicyResultSummary object
    • aggregateTimeWindow string
    • confidenceInterval number
    • ipsEstimatorDenominator number
    • ipsEstimatorNumerator number
    • nonZeroProbability number
    • snipsEstimatorDenominator number
    • sumOfSquares number
    • timeStamp string

RankRequest

  • RankRequest object: Request a set of actions to be ranked by the Personalizer service.
    • actions required array: The set of actions the Personalizer service can pick from.
    • contextFeatures array: Features of the context used for Personalizer as a
      • items object
    • deferActivation boolean: Send false if it is certain the rewardActionId in rank results will be shown to the user, therefore
    • eventId string: Optionally pass an eventId that uniquely identifies this Rank event.
    • excludedActions array: The set of action ids to exclude from ranking.
      • items string

RankResponse

  • RankResponse object: Returns which action to use as rewardActionId, and additional information about each action as a result of a Rank request.
    • eventId string: The eventId for the round trip from request to response.
    • ranking array: The calculated ranking for the current request.
    • rewardActionId string: The action chosen by the Personalizer service. This is the action your application should display, and for which to report the reward. This might not be the

RankableAction

  • RankableAction object: An action with it's associated features used for ranking.
    • features required array: List of dictionaries containing features.
      • items object
    • id required string: Id of the action.

RankedAction

  • RankedAction object: A ranked action with its resulting probability.
    • id string: Id of the action
    • probability number: Probability of the action

RewardRequest

  • RewardRequest object: Reward given to a rank response.
    • value required number: Reward to be assigned to an action. Value should be between -1 and 1 inclusive.

ServiceConfiguration

  • ServiceConfiguration object: The configuration of the service.
    • defaultReward required number: The reward given if a reward is not received within the specified wait time.
    • explorationPercentage required number: The percentage of rank responses that will use exploration.
    • logMirrorEnabled boolean: Flag indicates whether log mirroring is enabled.
    • logMirrorSasUri string: Azure storage account container SAS URI for log mirroring.
    • logRetentionDays required integer: Number of days historical logs are to be maintained. -1 implies the logs will never be deleted.
    • modelExportFrequency required string: Personalizer will start using the most updated trained model for online ranks automatically every specified time period.
    • rewardAggregation required string: The function used to process rewards, if multiple reward scores are received before rewardWaitTime is over.
    • rewardWaitTime required string: The time span waited until a request is marked with the default reward.