‎Use AI Studio as a data dource in Listening dashboards

Use AI Studio as a data dource in Listening dashboards

Updated

2 years ago

This article describes how to plot the AI Studio dimensions to a Listening dashboard to get instant insights into your AI Project’s related data.

Sprinklr’s AI Studio allows you to test the AI models to your satisfaction and provide feedback to customize the models as per your requirements. Besides being a stand-alone product, AI Studio can also be used as a data source in a Listening dashboard. In this way, you can easily plot the AI Studio dimensions to a Listening dashboard to get instant insights into your AI Project’s related data.

After you successfully create a Listening dashboard, follow these steps to use AI Studio as a data source while creating widgets.

To add AI Studio as a data source in a widget

On a Listening dashboard, click the Add Widget button in the top right corner. If the dashboard has no widget, the Add Widget option will also appear in the center of the window.
On the Widget builder, enter the widget name and select AI Studio as a data source.
Select a visualization type for your widget. For example, Table.
Select the AI Studio’s metrics and dimensions that you want to plot on your widget. You can select multiple dimensions with at least one metric to plot data.
Click Add to Dashboard to add the widget to the Listening dashboard.

The widget having AI Studio’s dimensions plotted is successfully added to your Listening dashboard. You can use the same way to add multiple widgets having different AI Studio dimensions.

To plot a widget using AI Studio dimensions

To plot a widget using AI Studio metrics and dimension, follow these steps –

Go to the Create New Widget screen in a Listening dashboard.
Select AI Studio as the source for the widget.
Choose the relevant metrics and dimensions for the widget.
Confirm the preview and create widget.
Generate the widget based on the selected metrics and dimensions.

Metric Name	Description
Messages Validated	This is the number of predicted messages’ classifications validated by the user.
Messages Edited	This is the number of predicted classifications in which the user switched the classification.
Messages Approved	This is the number of validations approved by the approver
Messages Rejected	This is the number of validations rejected by the approver
Insight Accuracy (PI)	Measures the proportion of correct Product insights predicted by the model
Message Sentiment Accuracy (PI)	Measures the proportion of correct Message Sentiments predicted by a PI validation model
Insight Accuracy (LI)	Measures the proportion of correct Listening insights predicted by the model
Message Sentiment Accuracy (LI)	Measures the proportion of correct Message Sentiments predicted by a LI validation model
Accuracy (Sentiment)	Accuracy for Sentiment project based on the number of sentiment insights approved and rejected

Dimension Name	Description
Conversation Stream	The conversations on social channels containing all messages that match the selected filters
User Group	User group as defined by the user
User	The name of the user
Project	The name of the project
Project Type	The type of the project created
Modified Date	The date when model was modified
Validated Date	The date when predictions were validated
Reviewed Date	The date when the classification was reviewed
Reviewed By	Name of the reviewer
Project Engine	The model engine of the project as selected while creation
Classification Final	The final classification of the message after it may have been changed
Classification Initial	The initial predicted classification of the message
Model Status	The status of the model
Project Created By	The user who created the product
Project Language	The language of the project added while creation
Project Status	The status of the project
Project Tag	Tags can be added to filter out specific projects in bulk
Text Classification Final	The final text classification of the message after it may have been changed
Text Classification Initial	The initial text classification of the message after it may have been changed