One of the essential tools of the Conversation Builder platform is Natural Language Understanding (NLU). This is what allows Intent Builder to analyze consumer input and assign accurate intents.
While LivePerson provides its own propriety NLU out of the box, the Conversation Builder also allows you to choose your preferred NLU Engine for analyzing text by routing all NLU analysis and training through an API.
This API layer of abstraction allows you to choose from the following NLU engines:
- LivePerson's native NLU
- Google Dialogflow
- IBM Watson
If you choose LivePerson's native NLU, no changes need to be made in Conversation Builder. This engine is already configured and set up by default.
LivePerson NLU supports intent detection for English and Spanish.
IBW Watson supports Arabic, Chinese, Dutch, English, French, German, Italian, Japanese, Korean, Portuguese, Russian, Spanish, and Swedish.
Google Dialogflow supports Cantonese, Chinese, Danish, Dutch, English, French, German, Hindi, Indonesian, Italian, Japanese, Korean, Norwegian, Polish, Portuguese, Russian, Spanish, Swedish, Thai, Turkish, and Ukranian.
Connecting a 3rd party NLU Engine
Step 1: Enable 3rd Party NLU Support
Contact your account administrator to enable your account for 3rd party NLU support.
Once 3rd party NLU support is enabled, you can start creating domains (with 3rd party NLU).
Step 2: Sign up and get the API keys
Register for or log in to an IBM Cloud account
Create or access a Watson Assistant resource
Generate Service Credentials with the role of Manager and an Auto Generated Service ID
View and copy the newly created credentials
Log in to the Dialogflow console
Create a new Dialogflow agent (which will create a new Google project)
Create a new Service Account for the newly created Google project with the role of Dialogflow API Admin
Create a JSON formatted private key for the service account by clicking Create key
View and copy the created key. This will be used in your dialogflow config data
Step 3: Create NLU Provider Credentials in Intent Builder
- Click on ‘NLU Provider Credentials’ on domain dashboard
- Create a new NLU provider credential
- Give a name for the credential
- Select the NLU Provider that you want to set the credential for
- Copy the credentials downloaded from Watson / Google Dialog Flow
Step 4: Add a domain for your NLU Provider
Import your intents and entities or add them later manually before Step 5.
Step 5: Train the domain
Click on the train button in the right icon menu
Select the NLU provider credential from the list in the train modal
Wait till the training is completed. Click on the refresh button to see the latest training status for the version.
Once training is completed, you can start testing with the modal version in the intent tester.
3rd Party NLU limitations
- Third Party NLU Domain length should not exceed 64 characters. (Watson limitation)
- Each domain can only support one language and it is available in the settings page.
NLU engine API limitations:
- Knowledge Base does not support 3rd party NLU right now
- Conversation Builder does not support pulling existing models from IBM Watson and Google Dialogflow into Intent Builder, only model push is supported.