A nice way to test your assumptions here is to pretend that you have perfect voice recognition of the above input already working, then figure out what software

A nice way to test your assumptions here is to pretend that you have perfect voice recognition of the above input already working, then figure out what software you need to write in order to answer all of the above questions. Does the tech to answer those questions (especially the ones with follow up and references) already exist? If so then this might become a reality soon. If you see some issues with parsing those sentences when they're input as strings then those issues will not magically go away with perfect speech recognition. Can you answer all those questions with state of the art software?