[NLP] expressions present in dataset not understood by chatbot

[former_member776464]

Hello,

I am finding recently that some expressions added to intentions in "train" are not being recognised when the expression is submitted to the chatbot by a user word by word.

After adding new expressions, the button train : has correctly been used and is green.

Does anyone have any insight as to how it is possible for the chatbot to not understand expressions which have been added to its data set ?

Thanks

[former_member523146]

Can you elaborate on what you mean by "not being recognized"? Do you mean that the corresponding intent wasn't assigned? Was it classified as a different intent?

How many intents do you have, how many expressions per intent?

[former_member776464]

Hi Stefan,

Here below some extra information :

- In "chat preview" the expression is not assigned to any intent and fallback is triggered

- In the "Expression analysis" window the expression is correcty assigned to the right intent

- The chatbot has about 60 intents, each with between 60 - 200 expressions. The intent which I have been testing has around 200 expressions

Thanks for your help !