Defining a classifier configuration

A classifier configuration defines connection information for an instance of IBM Watson® Discovery or IBM Watson Natural Language Understanding.

Before you begin

Configure the service that you want to connect to. For more information, see Configuring Watson Discovery and Configuring a Natural Language Understanding service on IBM Cloud. Make a note of the API key and the ID of the natural language classifier instance that you created.

To do this task, you need the Watson Mapping and Taxonomy Suggestions application permission.

Procedure

  1. Click Administration menu > Integrations > Mapping and Taxonomy Suggestions.
  2. Click New Classifier.
  3. Complete the fields in the Classifier Information section.
    1. Select Watson Discovery Analyze API or Natural Language Understanding in Watson Service Type.
    2. In Type, select Enumeration (taxonomy suggestions) or Association (object association suggestions).
    3. In Name, enter a name.
    4. In Description, enter descriptive information.
    5. Enter a link in Link. The link is optional and for information purposes only. The default is http://www.cloud.ibm.com.
  4. For a IBM Watson Discovery service, complete the fields in the IBM Watson Discovery Service section.
    1. Type your service URL in URL for Request. It is the URL to the IBM Watson Discovery Service on IBM Cloud Pak for Data that you configured.
    2. Type the Project ID.
    3. Type the Collection ID.
    4. Enter a Username.
    5. Enter a Password.
      If OpenPages and IBM Watson Discovery are both installed with the same IBM Cloud Pak for Data instance, you don't need to provide a password.
    6. Enter an API Version.
    7. Click Test Connection to verify the connection settings.
  5. For a Natural Language Understanding service, complete the fields in the IBM Watson Natural Language Understanding Service section. The person who trained the classifier model will have this information.
    1. Type your service URL in URL for Request. It is the URL to the Natural Language Understanding service on IBM Cloud.
    2. In API Key, enter the API key of the Natural Language Understanding instance.
    3. In Model ID, enter the identifier of the Natural Language Understanding classifier model.
    4. Click Test Connection to verify the connection settings.
  6. In the Usage Information section, in Confidence Threshold enter the lowest confidence score that a suggestion must meet.
  7. If Type is set to Enumeration, in the Field Settings section you can choose up to three fields. The order of the fields is important. Select the top-level field first, then the second-level field, and so on.
  8. If Type is set to Association, complete the Association section:
    1. In Object Type, select the object type that is associated, for example, SOXControl.
    2. In Association Type, select Child or Parent.
    3. In Object Path, enter a path to directories to search, for example, /Library.
    4. In Name Field, specify the field that is displayed as the object's name after a user selects an object to associate, for example, System Fields.Name. Format is <Field Group>.<Field Name>.
    5. In Description Field, specify the field that is displayed as the object's description after a user selects an object to associate, for example, System Fields.Description. Format is <Field Group>.<Field Name>.
  9. Click Create.

What to do next

Complete the remaining tasks that are described in Configuration overview for natural language classifier services.