I'm new to Dialogflow and NLU/NLP in general. From the examples I've seen, it seems possible to train Dialogflow with just a few phrases to parse two locations as different parameters depending on context, e.g. using "from" to infer that it's supposed to be the origin and "to" that the user means the destination.
I'm wondering how you would do this when the context is harder to determine than just looking at the preposition in front of the location.
Specifically, would Dialogflow still be able to tell apart whether I intend to leave or arrive at time X if I train it with just a couple of sentences like these:
Intent arrive_at_X:
- "I want to arrive in New York April 14th, at 8 PM."
- "I need to be in London by 4 PM next Monday."
- ...
Intent leave_at_X:
- "I want to leave for New York tomorrow at 8 AM."
- "I want to fly to London on Friday at 10 AM."
- ...
Or is the variance in this case too high because of the many different ways the user might phrase their request? Is there another, more elegant way to allow for both arrival and departure time requests or is it best practice to ask the user explicitly for one or the other?