Duckling and mitie

Getting Started with Rasa
1

Hello everyone,

I am new to Rasa and so far I wanted to use mitie entitiy and intents, but also have duckling extractor. Is that possible?

I am using rasa full and duckling from docker images, and I configured my pipeline to be like this:

language : "en"
pipeline:
  - name: "nlp_mitie"
    model: "data/total_word_feature_extractor.dat"
  - name: "tokenizer_mitie"
  - name: "ner_mitie"
  - name: "ner_synonyms"
  - name: "intent_entity_featurizer_regex"
  - name: "intent_featurizer_mitie"
  - name: "intent_classifier_sklearn"
  - name: "ner_duckling_http"
    dimensions: ["amount-of-money", "distance", "duration", "email", "phone-number", "quantity", "temperature", "time", "url", "volume"]

My docker-compose.yml file for rasa services is this one:

version: '3.5'
services:
  nlu-duckling:
    image: rasa/duckling
    ports: 
      - "8000:8000"
  nlu-service:
    image: rasa/rasa_nlu:latest-full
    ports:
      - 5001:5000
    volumes:
      - ./nlu/projects:/app/projects
      - ./nlu/logs:/app/logs
      - ./nlu/data:/app/data
      - ./nlu/config.yml:/app/config.yml

So far it is not working as expected. If I docker-compose up, services run fine and I can query things like “Hello John, the meeting is at 5pm”. Duckling works as expected returning me 5pm as time dimension. But If I train my model (with rasa-demo.json, for example), only mitie entity extractor works, and duckling does not work anymore. Is there any way to keep both working and returning their extracted entities together?

2

Just posting a self answer. I figured out that I needed to add the duckling service url to the pipeline as well. This change to the config.yml file solved my problem:

language : "en"
pipeline:
  - name: "nlp_mitie"
    model: "data/total_word_feature_extractor.dat"
  - name: "tokenizer_mitie"
  - name: "ner_mitie"
  - name: "ner_synonyms"
  - name: "intent_entity_featurizer_regex"
  - name: "intent_featurizer_mitie"
  - name: "intent_classifier_sklearn"problem
  - name: "ner_duckling_http"
    url: "http://nlu-duckling:8000"
    dimensions: ["amount-of-money", "distance", "duration", "email", "phone-number", "quantity", "temperature", "time", "url", "volume"]
2 Likes