Cant get duckling entities for Swedsh Language

Rasa Open Source
1

Which is the proper way of linking Rasa NLU for swedish language? First I downloaded spaCy Swedish language model, and I tried to link it by doing:

user:~$ pip3 install sv_model-0.0.0.tar.gz && python3 -m spacy link sv_model sv

Then I train and launch the Rasa nlu model:

python3 -m rasa_nlu.train -c config.yml --data data/corpus/ -o models --project current --verbose

and

python3 -m rasa_nlu.server --path ./models

However, when I do:

curl 'localhost:5000/parse?q=Jag%vill%flyga%från%Paris%till%London%imorgon&project=current

I dont get any duckling entity resolved. My config file looks like this:

language: 'sv'
pipeline: 
  - name: "nlp_spacy"              
  - name: "tokenizer_spacy"        
  - name: "intent_entity_featurizer_regex"
  - name: "intent_featurizer_spacy"
  - name: "ner_spacy"
  - name: "ner_crf"            
  - name: "ner_synonyms"
  - name: "intent_classifier_sklearn"
  - name: "ner_duckling_http"
    url: "http://0.0.0.0:8000"
    locale: "en_UK"
    dimensions: ["time", "number"]
    timezone: "UK/London"

And duckling is running like this:

no port specified, defaulting to port 8000
Listening on http://0.0.0.0:8000

When I just curl to the duckling server, I actually have Swedish language support, for example:

$ curl -XPOST http://0.0.0.0:8000/parse --data 'locale=sv_SV&text=imorgon på åtta'

Then:

[{"body":"imorgon på åtta","start":0,"value":{"values":[{"value":"2018-10-19T08:00:00.000-07:00",
"grain":"hour",
"type":"value"}]
,"value":"2018-10-19T08:00:00.000-07:00",
"grain":"hour",
"type":"value"},
"end":15,
"dim":"time",
"latent":false}

Therefore, my question is how can I get duckling support for Swedish language? I think I am not connecting well the SpaCy model with duckling because everything works well if I try with a different language like English.

2

Duckling is independent of spacy but your pipeline is incorret, duckling locale should be swedish

language: 'sv'
pipeline: 
  - name: "nlp_spacy"              
  - name: "tokenizer_spacy"        
  - name: "intent_entity_featurizer_regex"
  - name: "intent_featurizer_spacy"
  - name: "ner_spacy"
  - name: "ner_crf"            
  - name: "ner_synonyms"
  - name: "intent_classifier_sklearn"
  - name: "ner_duckling_http"
    url: "http://0.0.0.0:8000"
    locale: "sv_SV"
    dimensions: ["time", "number"]
    timezone: "UK/London"
1 Like
3

Thanks for the help!