Hello Guys, Hoping some one can help.
I am new to Rasa so apologies if I am asking naive question. I am working on a restaurant chatbot.
Do we explicitly need to give training examples for each of the synonyms set? In that case, doesn’t it look like ‘hard-coding’?
Please consider this example -
I have declared synonym for the location ‘Kochi’ as ‘Cochin’. Similarly I have done so for location ‘Bangaluru’ as ‘Bangalore’. Now, my training examples include an example for banglore, so NLU layer is able to identify the intent and extract entity for bangaluru/Banglore.
When the text is ‘Kochi’ - it extracts intent and entity - works perfect. But when the text is ‘Cochin’ - it only extracts intent but not entity. (Please not that, my training examples does not include example of Kochi).
Should I start giving examples for each of the synonym set? Why does NLU layer works for Kochi but not for cochin?
Note that - I have not included any example of Kochi itself, but it works perfect - intent and entity are extracted and confidence factor is - 0.96 for both.
thanks in advance guys!