Training data generation for FAQs answering bot

I am supposed to develop a chatbot for an elearning website that answers the user’s queries. It seems really hard to add intents for all the possible questions. If for instance I have a 1000 intents Do I need to add intents all of them? Is there any alternative to convert a dataset in excel form or .csv to RASA NLU suitable form. for Example If I have the dataset in excel with user’s text,intent,entity and value as the columns can we convert it into a dataset for RASA NLU. @Juste @akelad @harshitazilen @juste

I believe it will be challenging to support 1000 intents because confusion will increase with increase in number of intents (as there will be very thin line between near by intents). You can write a python script to convert your excel sheet or .csv file into Rasa acceptable JSON or MD file format. You can also take help ofRasaUI Tool to manage your training data.

1 Like

Thanks a ton! Will try doing it

There is a script for converting csv to nlu.md

1 Like