This is a core task in Natural Language Processing (NLP). When it comes to extracting entities from text, Hugging Face NER has become a popular and effective tool. Named Entity Recognition (NER) is the process of finding and classifying “entities” in text—like people’s names, company names, or locations.
This is how search engines understand that “Apple” in “Apple is releasing a new phone” is an Organization, not a Fruit.
Step 1: Installation
pip install transformers torch
Step 2: The Code
We use the token-classification pipeline. It’s called this because it classifies each word (or token).
from transformers import pipeline
# 1. Load the pipeline
# We'll use a popular, powerful NER model
ner_pipeline = pipeline(
"token-classification",
model="dslim/bert-base-NER",
grouped_entities=True # This is a helper to combine "Guido", "van", "Rossum"
)
# 2. Define your text
text = "My name is Alice, I live in New York, and I work for Google."
# 3. Run the NER model!
results = ner_pipeline(text)
# 4. Print the results
print("--- Entities Found ---")
for entity in results:
print(f"Text: {entity['word']}")
print(f"Type: {entity['entity_group']} ({entity['score']:.4f})")
print("-----")Step 3: The Result
The model will return a list of all the entities it found and their types:
--- Entities Found --- Text: Alice Type: PER (0.9982) ----- Text: New York Type: LOC (0.9990) ----- Text: Google Type: ORG (0.9987) -----
The model correctly identified Alice as a PERson, New York as a LOCation, and Google as an ORGanization.
Key Takeaways
- Named Entity Recognition (NER) finds and classifies entities like names and locations in text.
- Search engines distinguish entities, like interpreting ‘Apple’ as an Organization not a Fruit.
- The Hugging Face NER uses a token-classification pipeline to classify each word.
- The model provides a list of identified entities and their types, like Alice as a PERson and New York as a LOCation.
