chapter six

6 How to train your assistant

 

This chapter covers:

  • Training an assistant to recognize intents from user input
  • Collecting a viable training data set for your virtual assistant
  • Analyzing how the volume, variety, and veracity of training data affect an assistant’s performance

Training is an important part of building a virtual assistant.  Good training increases the intelligence of the assistant and enables it to be more useful to your users.  Training influences almost every single success metric and is important no matter which type of virtual assistant you are building.  This chapter will teach you what you need to start training your virtual assistant! 

FICTITIOUS INC prototyped their first virtual assistant using sample data from their virtual assistant provider.  This data trained the assistant to be proficient at recognizing generic customer service intents.  But what if a suitable sample data set did not exist?  How could FICTITIOUS INC build a training data set and use it to teach their virtual assistant?  Let’s explore a scenario where FICTITIOUS INC was not given a pre-created training set and had to build their own.

6.1       How to train a virtual assistant

6.2       How do you find training data?

6.2.1   Production logs

6.2.2   A mock user interface

6.2.3   Subject matter experts

6.2.4   Organizing training data into intents

6.3       Assessing if you have the right training data

6.3.1   Training data variety

6.3.2   Training data volume

6.3.3   Training data veracity

6.4       Summary