⚡️ Saturday AI Sparks 🤖 - 🔎 Named Entity Recognition (NER) with Transformers

Named Entity Recognition (NER) is a core task in Natural Language Processing (NLP). It automatically identifies and classifies parts of text into categories like persons, organizations, locations, and dates.

With Hugging Face Transformers, we can run state-of-the-art NER models in just a few lines of code — no training required.

Why NER?

• Information extraction → convert unstructured text into structured insights.
• Search & retrieval → improve indexing by tagging entities.
• Business use cases → extract names, places, and dates from contracts, emails, or support tickets.

Installing Requirements

pip install transformers torch

Loading a Pretrained NER Pipeline

We use Hugging Face’s dslim/bert-base-NER model, which is fine-tuned for entity recognition.

from transformers import pipeline

ner = pipeline("ner", model="dslim/bert-base-NER", aggregation_strategy="simple")

Here, aggregation_strategy="simple" groups together tokens that belong to the same entity.

Running NER on Text

Pass in a sample sentence containing organizations, people, and dates.

text = "Apple is looking to hire AI engineers in San Francisco starting in 2025. Tim Cook announced the plan."
entities = ner(text)

The output contains each entity, its predicted category, and a confidence score.

Sample Output

Apple → ORG (score: 1.00)
AI → MISC (score: 0.77)
San Francisco → LOC (score: 1.00)
Tim Cook → PER (score: 1.00)

Key Takeaways

• NER helps transform raw text into structured knowledge.
• Hugging Face makes it simple to apply pretrained models without training.
• This technique is widely used in chatbots, document parsing, compliance, and analytics pipelines.