Configure Natural Language Understanding in Virtual Agent

Configure Natural Language Understanding (NLU) in Virtual Agent to identify the NLU service provider for your instance. You can also specify the languages of NLU models used during conversation design, based on the languages supported by your NLU provider and the ServiceNow AI Platform®.

Before you begin

You can select only one NLU service provider for your instance. Virtual Agent supports the following NLU services:

• ServiceNow NLU (default): Create your models, intents, and entities in Virtual Agent Designer and NLU Workbench.
• IBM Watson Assistant: Configure the IBM Watson Assistant intent and entity integration so that Virtual Agent can access IBM Watson Assistant NLU model information.
• Microsoft LUIS: Configure the Microsoft LUIS Intent and Entity integration so that Virtual Agent can access Microsoft LUIS NLU model information.
• Google DialogFlow Essentials (ES): Configure the Google DialogFlow ES intent and entity integration so that Virtual Agent can access Google Dialogflow ES NLU model information.

The Activate toggle switch for NLU is toggled on for all new Virtual Agent activations and the NLU Service Provider defaults to ServiceNow NLU. The Activate toggle switch is toggled off for Virtual Agent Lite. If you upgraded from the previous release, Virtual Agent retains the NLU settings and integration information for your NLU service provider.

Role required: virtual_agent_admin or admin

About this task

The NLU-enabled topics that you preview, create, or update in Virtual Agent Designer must use NLU models created in the NLU service that you select here. For example, if you specify ServiceNow as the service provider, you can view and access topics that use ServiceNow NLU model groups.

Similarly, to work on NLU topics in multiple languages, the languages must be enabled in your NLU service and added to your model groups. For more information, see Multilingual model management.

Procedure

1. Navigate to All > Conversational Interfaces > Settings.
2. Click Virtual Agent.
3. Under Natural Language Understanding (NLU), click View Settings.
4. On the form, fill in the fields.

   Table 1. Natural Language Understanding (NLU) form

   Field	Description
   NLU Service Provider	Available NLU service providers for intent and entity extraction. The Supported NLU Languages list is displayed. It shows all of the language plugins that have been installed. For ServiceNow NLU, English is enabled to true by default. This field is automatically set to ServiceNow NLU.
   Ask user if topic VA chose is correct	Option to prompt the user to confirm that the matched intent is correct. This option enables user confirmation only on initial intent discovery. When enabled, Virtual Agent asks the user for confirmation before continuing to the automatically selected topic content. Enabling this field triggers the Intent confirmation message field. Figure 1. User confirmation example If the user selects No for an initial intent, then Virtual Agent runs the global fallback topic. If the user selects No for an automatically selected intent in mid-topic, then Virtual Agent returns to the topic that the user was in.
   Ask user if switched topic is correct	Option to prompt the user to confirm that the switched intent is correct. This option is enabled by default and prompts user confirmation only when the user triggers an intent switch in mid-topic. When enabled, Virtual Agent asks the user for confirmation before continuing to the switched topic content. Enabling this field triggers the Intent confirmation message field.

5. Edit the Intent confirmation message field, if needed.

   Text entered in this field is shown to the user to confirm that the matched intent is correct.

6. If you plan to use language-specific NLU models, enable the languages in the Supported NLU Languages list.
   A language is enabled if the Enabled column displays true. For more information, see Enable NLU languages in Virtual Agent settings.
7. Click Save.

Result

When you create or update a topic in Virtual Agent Designer, you can choose NLU models available for your specified NLU provider and the specified languages, if applicable. You can also set the NLU entity properties for the input controls that you add to your topic flow.

• Configure IBM Watson Assistant as the NLU provider for Virtual Agent
  Use the intents, entities, and utterances defined in IBM Watson Assistant and apply them as an NLU model for your Virtual Agent conversations.
• Configure Microsoft LUIS as the NLU provider for Virtual Agent
  Use the intents, entities, and utterances defined in a Microsoft Language Understanding Intelligent Service (LUIS) application and apply them as an NLU model for your Virtual Agent conversations.
• Configure Google Dialogflow ES as the NLU provider for Virtual Agent
  Use the intents, entities, and utterances defined in Google Dialogflow ES. Apply them as an NLU model for your Virtual Agent conversations.
• Enhanced natural language handling of user inputs in NLU topic discovery
  Give Virtual Agent the capability of recognizing natural variations on yes and no from the user, such as sure or nope. By defining a range of boolean responses as affirmative or negative, Virtual Agent doesn't have to prompt the user to restate their preference.