Automated Creation of Training Material
Using Natural Language Classification, your short tail classifier (be that Humley's Octavo or a third party provider) will use training data to match an intent to a user utterance, in order to provide a response.
To create this mapping, the classifier needs to be 'trained'.
To increase the accuracy of your conversational assistant, it's recommended to create as much diverse training material, i.e. Alternative User Inputs, as possible to give the classifier the best chance to make high confidence (and therefore, first-time) matches. This can require a lot of time and resource. Which is where Octavo comes in.
When enabled, Octavo will study each intent and automatically create Alternative User Inputs for them, each time that the knowledge base is trained.
See more on how to Train Intents
The Octavo page has two sections -
Automated Training Material
Enable or disable the automated creation of training material with the toggle provided
This will also cause the On/Off toggle on the Train message box to default to the same, although this can be changed there as desired
Alternative Input Generated
Shows how many Alternative Inputs have been automatically generated
N.B. These are stored against each intent, but are not currently visible
The last date & time training occured with the Automated Generation enabled
Octavo Time Saved
An estimation of the time saved which otherwise would have been needed to create the Alternative User Inputs generated
- Delete All Octavo training material
Click to delete all training material generated by Octavo