I am training Bixby to recognize hashtags and do something different if an utterance begins with a hashtag than if it begins with something else. So far I have provided training examples in the format
hashtag word
hashtag word word
hashtag word word word
Should I also provide examples in formats like
hashtag number word word
For example,
hashtag 420 From Home
Another way of stating this is that I am asking does the training engine treat "words", numbers, and alphanumeric strings as being of equal weight when generating rules.
As a general rule, your training should include representative samples of what you expect the user to say. But it is more important to get the pattern of the utterance correct.
The following utterances in your training will inform Bixby to expect 1, 2, or many words after hashtag.
w1, w2, w3 can be either "alphanumeric" or words or numbers.