Confidence Score / Confidence Level

-> to BOTwiki - The Chatbot Wiki

The Confidence Score, or Classification Threshhold, indicates how certain the NLP Service / Machine Learning Model is that the respective intent has been assigned correctly. The score can have a value between 0 and 1, depending on how the neural networks work. In general, a score is calculated for each intent for each user input and the one with the highest value is returned as the result. If the confidence level falls below a pre-defined limit, a Fallback Intent is output.

An example for the calculation of the Confidence Score at Google Dialogflow:

For the Intent "Burger_bestellen", the following four training phrases in the system:

"Burger order."

"Order burgers."

"I want to order a burger."

"I want to order a burger"

When the user entered "I'd love to order a burger from you", the following was displayed Dialogflow a confidence score of 0.8. The NLP Service Dialogflow 80 % is therefore sure that the response output from the intent "Order Burger" was correct. Due to this data situation and on the basis of previously defined rules, no fallback intent is now output.

> Back to the BOTwiki - The Chatbot Wiki