Use patterns to differentiate similar utterances

• 2 minutes

In some cases, a model might contain multiple intents for which utterances are likely to be similar. You can use the pattern of utterances to disambiguate the intents while minimizing the number of sample utterances.

For example, consider the following utterances:

• "Turn on the kitchen light"
• "Is the kitchen light on?"
• "Turn off the kitchen light"

These utterances are syntactically similar, with only a few differences in words or punctuation. However, they represent three different intents (which could be named TurnOnDevice, GetDeviceStatus, and TurnOffDevice). Additionally, the intents could apply to a wide range of entity values. In addition to "kitchen light", the intent could apply to "living room light", television", or any other device that the model might need to support.

To correctly train your model, provide a handful of examples of each intent that specify the different formats of utterances.

• TurnOnDevice:
  • "Turn on the {DeviceName}"
  • "Switch on the {DeviceName}"
  • "Turn the {DeviceName} on"
• GetDeviceStatus:
  • "Is the {DeviceName} on[?]"
• TurnOffDevice:
  • "Turn the {DeviceName} off"
  • "Switch off the {DeviceName}"
  • "Turn off the {DeviceName}"

When you teach your model with each different type of utterance, the Azure AI Language service can learn how to categorize intents correctly based off format and punctuation.