Skip to main content
Do you like Artifex? Give it a ⭐ star on GitHub!

🏷️ Named Entity Recognition model

Use the default Named Entity Recognition model​

Need a general-purpose Named Entity Recognition model? You can use Artifex's default Named Entity Recognition model, which is trained to recognize fourteen named entities out-of-the-box:

  • PERSON
  • ORG
  • LOCATION
  • DATE
  • TIME
  • PERCENT
  • NUMBER
  • FACILITY
  • PRODUCT
  • WORK_OF_ART
  • LANGUAGE
  • NORP (Nationalities or religious/political groups)
  • ADDRESS
  • PHONE_NUMBER
from artifex import Artifex

ner = Artifex().named_entity_recognition

print(ner("John landed in Barcelona at 15:45."))

# >>> [[{'entity_group': 'PERSON', 'score': 0.92174554, 'word': 'John', 'start': 0, 'end': 4}, {'entity_group': 'LOCATION', 'score': 0.9853817, 'word': ' Barcelona', 'start': 15, 'end': 24}, {'entity_group': 'TIME', 'score': 0.98645407, 'word': ' 15:45.', 'start': 28, 'end': 34}]]

Learn more about the default Named Entity Recognition model on our Named Entity Recognition HF model page.

Create & use a custom Named Entity Recognition model​

Need more control over the named entities recognized, or do you want to tailor the model to your specific domain for better results? Fine-tune your own Named Entity Recognition model, use it locally on CPU and keep it forever:

from artifex import Artifex

ner = Artifex().named_entity_recognition

model_output_path = "./output_model/"

ner.train(
    domain="medical documents", # change to your desired domain
    named_entities={ # define your custom named entities
        "PERSON": "Individual people, fictional characters",
        "ORG": "Companies, institutions, agencies",
        "LOCATION": "Geographical areas",
        "DATE": "Absolute or relative dates, including years, months and/or days",
        "TIME": "Specific time of the day",
        "NUMBER": "Numeric measurements or expressions",
        "WORK_OF_ART": "Titles of creative works",
        "LANGUAGE": "Natural or programming languages",
        "NORP": "National, religious or political groups",
        "ADDRESS": "full addresses",
        "PHONE_NUMBER": "telephone numbers",
    },
    output_path=model_output_path,
)

ner.load(model_output_path)
print(ner("The patient John Doe visited New York on 12th March 2023 at 10:30 AM."))

# >>> [[{'entity_group': 'PERSON', 'score': 0.9456123, 'word': 'John Doe', 'start': 12, 'end': 20}, {'entity_group': 'LOCATION', 'score': 0.9783456, 'word': ' New York', 'start': 29, 'end': 38}, {'entity_group': 'DATE', 'score': 0.9654321, 'word': ' 12th March 2023', 'start': 42, 'end': 58}, {'entity_group': 'TIME', 'score': 0.9543210, 'word': ' 10:30 AM.', 'start': 62, 'end': 71}]]