Closed samuelvkwong closed 1 year ago
As we discussed before, I would see no problem in extending the modeling language to allow for such behaviour. The existing specification woul coexist with the new one giving the bot developers the option to choose between the two.
@AlexanderNeumann Do you have any thoughts or example usages for this implementation?
I agree with Ben's comment.
I think Ben has already examples from his thesis. @lakhoune can you provide a concrete example of the mensa bot for example?
From use case scenarios, there are plenty. For example, if the response of a request (when we fetch information from a service/website etc) is relevant to the bot message. If the request fails the state should not change or go into another state (automatically). It is also a little bit complicated because sometimes we are using webhooks if the requests are taking longer. We first directly give a response to the user that there is a longer process ongoing and when the process is finished we give the result to the user.
An example for sequential behaviour for the mensabot would be the review process.
The process looks similar to this:
'request to make review' -> 'specify canteen' -> 'specify meal' -> 'confirm choices' -> 'give stars' -> '(optional) feedback'
Currently we do this in one state for the first half 'request to make review' -> 'specify canteen' -> 'specify meal' -> 'confirm choices'
and then a second half 'give stars' -> '(optional) feedback'
But it would be great to have each as an own bot action