Using confidence scores to improve conversation accuracy

There are two steps that Amazon Lex V2 uses to determine what a user says. The first, automatic speech recognition (ASR), creates a transcript of the user's audio utterance. The second, natural language understanding (NLU), determines the meaning of the user's utterance to recognize the user's intent or the value of slots.

By default, Amazon Lex V2 returns the most likely result from ASR and NLU. At times it may be difficult for Amazon Lex V2 to determine the most likely result. In that case, it returns several possible results along with a confidence score that indicates how likely the result is correct. A confidence score is a rating that Amazon Lex V2 provides that shows the relative confidence that it has in the result. Confidence scores range from 0.0 to 1.0.

You can use your domain knowledge with the confidence score to help determine the correct interpretation of the ASR or NLU result.

The ASR, or transcription, confidence score is a rating on how confident Amazon Lex V2 is that a particular transcription is correct. The NLU, or intent, confidence score is a rating on how confident Amazon Lex V2 is that the intent specified by the top transcription is correct. Use the confidence score that best fits your application.