If I am using below data to train , it’s memorizing the entity values(when not using roles) or memorizing the roles(when using roles) , so its not filling the slots dynamically.
nlu.yml:
- intent: book_train
examples: |
- Please purchase train ticket for me
- I want to book ticket from [Delhi]{“entity”: “location”, “role”: “source”} to [Kolkata]{“entity”: “location”, “role”: “destination”}
- I want to buy ticket
- Book train ticket
- Please book a ticket to [Bangalore]{“entity”: “location”, “role”: “destination”}
- please buy ticket from [Kolkata]{“entity”: “location”, “role”: “source”} to [Chennai]{“entity”: “location”, “role”: “destination”}
- I want a flight from [Assam]{“entity”: “location”, “role”: “source”} to [Sikkim]{“entity”: “location”, “role”: “destination”}
Below is the test confidence:
{
“text”: “Book a ticket from Kolkata”,
“intent”: {
“name”: “book_train”,
“confidence”: 0.9998188614845276
},
“entities”: [
{
“entity”: “location”,
“start”: 19,
“end”: 26,
“confidence_entity”: 0.9973909854888916,
“role”: “destination”,
“confidence_role”: 0.6011223196983337,
“value”: “Kolkata”,
“extractor”: “DIETClassifier”
}
],
Kindly help me on this