Entity Extraction In Urdu

I am building a chatbot in Urdu language and I was working on my entity extraction. The problem that I have is that when an entity e.g. “location” has a value in the training data, the slot is filled correctly. But when the entity value is not in the training set, I get the error:

Failed to extract slot location with action emergency_form

my config.yml is provided below:

language: en
pipeline:

  • name: WhitespaceTokenizer
  • name: CRFEntityExtractor
  • name: EntitySynonymMapper
  • name: CountVectorsFeaturizer token_pattern: (?u)\b\w+\b
  • name: EmbeddingIntentClassifier

policies:

  • name: FallbackPolicy
  • name: MemoizationPolicy
  • name: FormPolicy
  • name: MappingPolicy

is there a way to solve this issue?

You mean you have an overfitting problem?

I don’t think it overfits. I gave 20 examples for the entity location. Only those examples are correctly recognized and not new ones. Is there something that I might be doing wrong.