Creating a text categorization model to run topic detection models in Cloud AutoML

To run the topic detection models that you create in Cloud AutoML, Google's cloud-based machine learning service, configure a text categorization model in Prediction Studio using the machine learning service connection to Cloud AutoML.

Before you begin: Define your model, as well as the machine learning service connection:
  1. In Google AutoML, create a topic detection model.
  2. In Dev Studio, connect to your machine learning service instance by creating an OAuth 2.0 authentication profile.

    For more information, see Creating an authentication profile.

  3. In Prediction Studio, define your machine learning service connection.

    For more information, see Configuring a machine learning service connection.

  1. In the navigation pane of Prediction Studio, click Models.
  2. In the header of the Models work area, click New > Text categorization.
  3. In the New text categorization model window, set up your topic detection model:
    1. In the New model name field, enter a unique name for your model.
    2. In the Save to channel list, select the channel to which you want to save your model, for example, a chatbot channel.
    3. In the Apply to field, specify the class to which you want to save the model, and then specify its ruleset or branch.
    4. In the Detection section, select Topics.
    5. In the Text analytics service list, select Google AutoML NLP.
    6. In the Language list, select the language for the model to use.
    7. In the Service name list, select the machine learning service that you defined in step 3 in the Before you begin section.
    8. In the Model identifier list, select the model that you want to connect to.
    9. Optional: To view the topics that the model detects, in the Topics section, click View.
    10. Click Create.
  4. In the Text Categorization - Topic Model workspace, review the model settings.
  5. Optional: To test the model, in the Test the model section, in the Sample text field, enter some sample text, and then click Test to check that the created model detects the topic.
  6. Click Save.