Lex Bot is not working correctly

0

we are facing issue in lex bot as when we provide the described uttarances as a voice to lex bot it doesn't take the correct intent and enters in wrong intent.

  • Can anyone help us with above problem as we are sort of time.

asked 6 months ago286 views
2 Answers
0

There are a few things you can try:

  1. Review the best practices for creating your interaction model and see if there are improvements that could be made to your bot based on those. Be aware that Lex tunes the transcription per bot based on the sample utterances you have given, so adjusting them can help both transcription and intent recognition.
  2. Turn on conversation logs. These will help you identify issues. Look for a some of the following:
  • Transcription errors, or low transcription confidence. You can mitigate these by ensuring you have the best language selected from Lex, and also following the best practices for your sample utterances and slot values.
  • Low nlu confidence scores, or intents showing very close confidence scores. The logs will show some of the other intents that were considered and their confidence scores. If some are very close in confidence scores then it it is likely the sample utterances are too close in those.

If you can narrow down the issue and give more specifics there are other features that could help as well, but without finding the underlying cause it would be hard to give an exact recommendation.

answered 6 months ago
-1

In general, when the Lex service fails to match a user's utterance with the correct intent for slots, it suggests that the sample utterances do not cover a wide enough range of user inputs. As a best practice, the sample utterances specified for intents and slots should closely match your expected user input. Lex can learn from previous user inputs, so it's beneficial to test your bot with a variety of voice inputs over time.

That being said, it is also possible that the issue lies with the Lex service's inability to understand speech patterns. This is a different problem from the one mentioned above. Lex utilizes AWS Comprehend to translate user speech into text that Lex can understand and correlate with your bot's intents. You can enable audio logs to understand how Lex is interpreting user voice patterns; for more details, please refer to Audio Logs Configuration documentation below. https://docs.aws.amazon.com/lexv2/latest/dg/conversation-logs-configure.html

Some other potential issues include the user's utterances not matching the slot values for your intent. You can also check the logs to see if this is the case. If these general suggestions do not resolve your issue, especially if the problem lies with Lex's interpretation of user speech, I would recommend reaching out to Premium Support via a support case for further assistance.

odwa_y
answered 6 months ago

You are not logged in. Log in to post an answer.

A good answer clearly answers the question and provides constructive feedback and encourages professional growth in the question asker.

Guidelines for Answering Questions