Hello Rasa team,
I was wondering how to have a better experience with a questionnaire answered using a Likert scale and the bot can also receive open responses. For example, U: start form B: How do you find this bot?
- Excellent
- Very Good
- Good
- Fair
- Poor
U: this bot is excellent. It offers a lot of things and can answer quite well.
In the above example, my base bot saves the whole response as text to a slot. May I ask for advice if I want something like…
full_response1: “this bot is excellent. It offers a lot of things and can answer quite well.” response1: excellent
Why get full_response1? So that it can be part of the training data later on. Why response1 is necessary? To get an immediate score from the bot.
Another approach that I am thinking of is with multiple turns i.e., ask first exact response from the scale. Then, ask another question for an explanation. However, with even 3-5 questions, this will seem too long.