Used backend / pipeline (spacy_sklearn.):
Operating system (UBUNTU):
Issue:
I want an example of using duckling in rasa NLU.
Content of configuration file (if used & relevant):
I answered this on Gitter, are the docs insufficient?
https://rasa-nlu.readthedocs.io/en/latest/pipeline.html#ner-duckling
In short:
ner_duckling or ner_duckling_http respectively to above)@dilipbobby did that help?
for now I'm clear :) thank you .if I have any problem I will reopen the issue
will ner_duckling classify any named entity a user specifies, or do I need to add training data with that entity?
Hello!
My current config file is in json format and I install duckling module but I don't know how to add it in the config file.This is the content of file(path and data are not relevant):
{
"project":"nav-os",
"language":"en",
"pipeline":"spacy_sklearn",
"path" :"/home/costi/Documents/arhive/automated-data-training-master",
"data" :"/home/costi/Documents/arhive/rasa_train.json"
}
I don't have acces to your Gitter.Thanks
which version of rasa_nlu are you using? this is a very old config format 😄
you should change your pipeline to contain all the specific components, and then add ner_duckling like specified here: https://nlu.rasa.com/pipeline.html#ner-duckling
Is duckling also supported for german? The wit.ai website does not list german:
https://duckling.wit.ai/
But the python wrapper does:
https://github.com/FraBle/python-duckling/blob/master/duckling/language.py#L8
Now I am wondering if german duckling is supported and the docs are just old. Did you test that?
it does indeed support it
@akelad Is there any way to use wit.ai duckling component in rasa_nlu? or is they are same?
It is the same. Keep in mind though that ner_duckling is being phased out and uses a very old version of Wit's Duckling. It's recommended to use ner_duckling_http, which can use the latest version of Duckling.
@wrathagom unlike ner_duckling, ner_duckling_http doesn't work at all. Can you help me, why? Here is my pipeline
pipeline:
- name: "nlp_spacy"
- name: "tokenizer_spacy"
- name: "intent_entity_featurizer_regex"
- name: "intent_featurizer_spacy"
- name: "ner_crf"
- name: "ner_duckling_http"
- name: "ner_synonyms"
- name: "intent_classifier_sklearn"
please ask usage questions on the forum, there's info on how to configure this here: https://rasa.com/docs/nlu/components/#ner-duckling-http
@wrathagom can I user 'ner_duckling_http' and 'ner_cnf' both in the configuration file or 'ner_duckling_http' is enough?
@akelad - the sad part is the urls you have specified seems to be not working now. We keep getting "Oops… this page doesn't exist." error when we navigate.
Most helpful comment
It is the same. Keep in mind though that
ner_ducklingis being phased out and uses a very old version of Wit's Duckling. It's recommended to usener_duckling_http, which can use the latest version of Duckling.