Blindset testing

60 minutes

• Introduction into using tags

  • Using tags at the intent level
  • Using tags at the utterance level

• Running a blindset

  • Reviewing what a blindset is and when to use one
  • Utilising label names as an example
  • Adding the literal intent name to every intent from a csv
  • Understanding why is this a good test

• Understanding what you want to achieve with blindsets

  • The different types of blindsets
    • Regression - middle of the meaning, cases that should match
    • Analyst - difficult or edge cases
    • Specific sets for specific purposes

• Configuring the HumanFirst to exclude blinds by intent tags

• Configuring HumanFirst to run an evaluation on an external NLU

• Importing a blindset from a CSV and automatically adding a phrase level tag

• Blindsets as a way of having a structured empircal approaches for redesigning classes