I have been training a bot that uses a significant number of different entity types. My training set contains many examples and I am also looking at using lookup tables to provide additional values.
My entity dictionaries and quite large in size(some in the millions) and can change quite often, so I would prefer not to create massive lookup tables, and retrain the model too often.
I have an existing entity recognition system that works very well and would like to feed that into the intent interpreter somehow.
For an example query how do I get to Piccadilly Circus, I have already identified Piccadilly Circus as the place entity and would like it to match the intent that I have trained with how do I get to [Westminster Abbey](place)
If anyone can give any direction as to if and how this could be achieved, through custom components or otherwise, that would be great.
Thanks, Chris