Amazon Lex launches accuracy improvements and confidence scores

Posted on: Aug 6, 2020

Amazon Lex is a service for building conversational interfaces into any application using voice and text. Today, we are launching natural language understanding improvements and confidence scores on Amazon Lex. We continuously improve the service based on customer feedback and advances in research. The natural language understanding improvements enable better intent classification accuracy. Confidence scores indicate the likelihood of a certain intent and can be used to enhance conversation design.

Confidence scores are surfaced for the top five intents in the bot. You can combine confidence scores with business knowledge to improve your understanding of a user’s intent. Consider a customer’s request to a banking bot, “What’s my balance?” In this example, the bot identifies multiple intents (checking, saving, or credit card accounts) which reflects the ambiguity of the request. In such a case where two or more intents are matched with reasonably high confidence, intent classification confidence scores can help you determine when you need to use business logic to clarify the user’s intent. If the user only has a credit card then you can trigger the intent to surface the balance on the credit card. Alternately, if the user has both a credit card and a checking account, you can pose a clarification question such as “Is that on your credit card or checking account?” before proceeding with the query. You now have better insights to manage the conversation flow and create more effective conversations.

You can opt in and enable accuracy improvements and confidence scores via the Console or SDK in the N. Virginia, Oregon, Sydney, and Dublin regions. For customers in London, Frankfurt, Tokyo, and Singapore regions, these features are available by default. Please visit documentation for more details.