Open alberchou opened 1 year ago
Other thing, that can be a bug (maybe): When I put a large list of values (that are previously added with nlp.addNamedEntityText function), the number of coincidences are less than the passed in the original string (in my case: 22 entities were found while sent a text chain of 29 values).
I have some entities with dots inside (for example: aaaaa.bbbbb.ccccc) and I need to set 1 as accuracy but if I do that those entities are not recognized,
Is there any option to exclude some characters from being used as token separator? Furthermore, is it possible to use that only for entity recognition (not for intent recognition).
Thank you!