I am using spacy for number identification.
Configuration for spacy is [ “ORDINAL”] .
It works well when my query text contain number like
1,00,000
but when I asked number like
19000000
in my query text which failed to identify number.
Regex will help in this case. See the regex part in the documentation. https://rasa.com/docs/nlu/0.13.8/dataformat/#section-dataformat
Thanks
I added:
“regex_feature”:[{“pattern”: “\d+”, “name”: “number”}]
to my training file and is working great for any number in a sentence that is being recongnized by the intent classifier (otherwise it isn’t recognized, for example, I tried with a phrase consisting only on numbers and I have no examples like those in my training set).
Funny thing… I trained after adding the regex and at first i didn’t seemed to work. I just tried again and is working ok… just to keep in mind, maybe something in cache… I don’t know