Rasa: duckling entities not generating

[praveenrengarajan]

Rasa Core version: 0.8

Python version: 3.6

Operating system (windows, osx, ...): Windows 10 64

Issue:
@mouadino
nlu_config.json
{ "pipeline": [ "nlp_spacy", "spacy_sklearn", "tokenizer_spacy", "intent_featurizer_spacy", "intent_classifier_sklearn", "ner_crf", "ner_synonyms", "ner_duckling" ], "path" : "./models/nlu", "data" : "./data/training_data.json", "duckling_dimensions": ["time", "number", "amount-of-money", "distance" , "email" , "url", "phone-number", "duration"] }
Its creating ner_duckling.json file in model folder
{ "dimensions": [ "time", "number", "amount-of-money", "distance", "email", "url", "phone-number", "duration" ] }
but in training_data.json its not giving duckling entities
{ "intent": "info", "entities": [], "text": "the first Tuesday of October" }

[praveenrengarajan]

@mouadino

[akelad]

training_data.json won't contain any entities from duckling. That's fine though, this is training data you've written yourself. Ner_duckling will predict entities when you've actually got the model loaded.

[praveenrengarajan]

It's working after I declared slot name as time and date. but one thing how can I declare that in stories.

time:{to:'2018..',from: '2018..'} and in some other scenario only time: '2018...' @akelad

[mouadino]

@praveenrengarajan We did that using interactive learning, with that the slot get generated automatically you can check our examples from the gist here https://gist.github.com/mouadino/78ab561e9164b4857528745e6668c0f1