Interesting, a good description of methods, which are already out there, on the chatbot description books, and perhaps in the framework manuals. ;)
I would have been delighted to read some kind of evaluation on each of the 3 methods you mention, (there are also others, as well as combinations)
I like to read some kind comments like:
“NLU is good for aaaa” but fails on XX languages, then
“NLP works fine for bbbb, but not so good for cccc”
Also there are many platforms out there using one or other methods, and the high language dependency on the aforementioned methods, are also non-despicable. If you can, please clarify and expand.
Also, if you might answer this: thank you, in advance!