Hi everyone!:slight_smile: I’ve started developing a chatbot but so far I’m only training/testing it locally on my machine. Since in the end it should be running on a server in order for users to interact with it I was wondering about the following question: Is there a built-in feature in rasa stack that allows to collect new input from users in order to increase the size of the training datasets for both the NLU and core unit in the Rasa stack over time? (Of course there would be a lot of manual (“hand-labelling”) work once the data is collected.) I think it should be realizable with a custom action: e.g. whenever the fallback policy is needed some custom code could store all data of interest. However, since I am quite new to chatBots, APIs and databases I’m sure I don’t see the “correct” way of doing this. Could someone please point me to the right direction? Specifically, how would I trigger the task of collecting a given conversation and how/where would I then store it? Is all this offered in the data API of Rasa platform? Thank you so much!:slight_smile:

How to collect new data from users' conversations

akelad (Akela Drissner) November 2, 2018, 3:20pm 2

yes it’s part of the platform! Feel free to contact our sales team if you’re interested

Topic		Replies	Views
Traing data for Rasa Tutorials, Resources & Videos	3	331	August 31, 2020
Training from chat data Rasa Open Source	1	306	August 3, 2020
Training core from user chat Rasa Open Source	1	523	December 20, 2018
How to make conversational chatbot with Rasa Rasa Open Source	2	498	March 22, 2022
Dynamic data set, is it possible to create rasa chatbot Rasa Open Source	1	763	February 11, 2020