Improving the Bot using Training
To improve the User experience with the Bot, it is required to train Luma VA periodically. It means training the NLP engine on how to understand the user's language better. The more phrases are fed to the Natural Language Processing (NLP) engine, the better it matches skills to the phrases and even matches to phrases that do not exist.
A Luma Skill is triggered when a user request matches user phrases associated with the skill. When the Virtual Agent does not find any matching skill for a user’s utterance, it is marked as 'Unmatched Phrase'. There are two ways to Train Luma VA on these unmatched phrases:
Supervised Training
Automated Training
Supervised Training
You can manually train your bot by identifying and associating the unmatched phrases to System Skills and Custom Skills available in your bot. To do so, follow the below steps:
Navigate to the Unmatched Phrases tab on the Supervised Training Page under the Bot menu.
Select one or more Unmatched Phrases as required.
Click on Assign to Skill.
Select the matching skill from the list and click Assign to Skill.
In case the phrase does not match any existing skill, you may make the Phrase as Reviewed.
Run Build and Publish for training to take effect in Virtual Agent.
Automated Training
Luma VA can also be enabled to train itself automatically, based on the phrases used by the End-user to execute a skill. It automatically associates an unmatched phrase to an existing skill without any manual intervention.
To enable Automated training for your bot, do the following:
Navigate to Bot → Configurations → NLP Settings → Disambiguation tab.
On Confirmation Settings, Toggle the Automate Training With User Responses and set it to Active.
Click Save.
To learn more about the conditions for Automated Training you can refer to Automated Training documentation.