I’m making a bot that is an AI FAQ of sorts, an information reference bot. A lot of questions could be written as “Tell me about the magical effects” or “Tell me about magical effects” (difference: a ‘the’ is added in the first example).
Should it be trained on both with and without the ‘the’, with [magical effects] as the entity, or would it be better to train it on “Tell me about [the magical effects]” and set a synonym for “the magical effects” and “magical effects”?
Or do both? Train with and without the ‘the’, and also set a synonym? Or would that confuse the issue?
Just not sure which route would result in a better/more robust model.