Module 10.7: Named Entity Recognition (NER)

Named Entity Recognition (NER) is a key technique in Natural Language Processing (NLP) that focuses on identifying and classifying important information in text into predefined categories such as names of people, organizations, locations, dates, and more.

NER helps machines understand “who”, “what”, “where”, and “when” from text data. It is widely used in search engines, chatbots, information extraction systems, and AI-powered applications.

In this tutorial, we will learn what NER is, how it works, types of entities, examples, techniques, workflow, advantages, limitations, and real-world applications in Artificial Intelligence systems.

What is Named Entity Recognition (NER)?

Named Entity Recognition is the process of identifying and classifying named entities in text into predefined categories.

Simple Definition

NER is a technique in NLP that extracts important real-world objects like names, places, organizations, and dates from text.

Why is NER Important?

Human language contains a lot of meaningful information hidden inside text. NER helps extract structured data from unstructured text.

Importance of NER

• Converts unstructured text into structured data.
• Improves information extraction.
• Enhances search engine results.
• Supports chatbots and virtual assistants.
• Helps in data analytics and business intelligence.

Types of Named Entities

NER systems typically identify the following entity types:

1. Person (PER)

Names of individuals.

Elon Musk, Sundar Pichai, Albert Einstein

2. Organization (ORG)

Names of companies, institutions, or groups.

Google, Microsoft, OpenAI

3. Location (LOC)

Names of places, cities, countries, landmarks.

India, Kolkata, Eiffel Tower

4. Date & Time (DATE/TIME)

References to time expressions.

2026, Monday, 5th May, tomorrow

5. Money (MONEY)

Monetary values.

$100, ₹5000, 1 million dollars

6. Percent (PERCENT)

Percentage values.

50%, 99.9%

Example of NER

Input Sentence

Elon Musk founded SpaceX in the United States in 2002.

NER Output

Elon Musk → PERSON
SpaceX → ORGANIZATION
United States → LOCATION
2002 → DATE

How NER Works

NER systems analyze text and classify words based on context, grammar, and machine learning models.

Workflow

Input Text
   ↓
Tokenization
   ↓
Feature Extraction
   ↓
Model Prediction
   ↓
Entity Classification
   ↓
Output Entities

Approaches to NER

1. Rule-Based Approach

Uses predefined rules, patterns, and dictionaries.

Example

• Capitalized words → possible names
• Words after titles like Mr., Dr. → person entity

2. Machine Learning Approach

Uses algorithms like CRF (Conditional Random Fields), SVM, etc.

3. Deep Learning Approach

Uses neural networks like LSTM, BiLSTM, and Transformers.

4. Transformer-Based Models

Modern NER systems use models like BERT for high accuracy.

NER in NLP Pipeline

Raw Text
   ↓
Preprocessing
   ↓
Tokenization
   ↓
NER Model
   ↓
Entity Extraction
   ↓
Structured Output

Example: Step-by-Step NER

Input Sentence

Google CEO Sundar Pichai announced a new AI project in India.

Step 1: Tokenization

Google | CEO | Sundar | Pichai | announced | a | new | AI | project | in | India

Step 2: Entity Detection

Google → ORGANIZATION
Sundar Pichai → PERSON
India → LOCATION

NER Tagging Formats

IOB Format

Common format used in NER labeling.

I-ORG → Inside Organization
I-PER → Inside Person
B-LOC → Beginning Location
O → Outside any entity

Applications of NER

1. Search Engines

Improves search accuracy by identifying key entities.

2. Chatbots

Helps extract important information from user queries.

3. News Classification

Automatically categorizes news based on entities.

4. Information Extraction

Extracts structured data from documents.

5. Customer Support

Identifies customer names, products, and issues.

6. Healthcare

Extracts disease names, drugs, and medical terms.

Example in Real Life

Sentence

Apple released the iPhone in California on September 12, 2026.

NER Output

Apple → ORGANIZATION
iPhone → PRODUCT
California → LOCATION
September 12, 2026 → DATE

Advantages of NER

• Extracts structured data from text.
• Improves AI understanding of language.
• Useful for automation systems.
• Enhances search engine results.
• Supports data analysis and business intelligence.

Limitations of NER

• Difficult with ambiguous names.
• Struggles with slang and informal text.
• Requires large training data.
• Language-dependent performance.
• May fail in complex sentences.

Challenges in NER

• Entity ambiguity (Apple = company or fruit)
• Context understanding
• Multilingual text processing
• New or unseen entities
• Noisy social media data

NER vs Other NLP Tasks

NER Workflow Summary

Input Text
   ↓
Tokenization
   ↓
Context Analysis
   ↓
Entity Detection
   ↓
Classification
   ↓
Final Output

Best Practices

• Use transformer-based models for accuracy.
• Train on domain-specific datasets.
• Handle ambiguous words carefully.
• Combine NER with POS tagging.
• Clean text before processing.

Key Terms to Remember

• Named Entity Recognition
• NER Tags
• Person Entity
• Organization Entity
• Location Entity
• IOB Format
• Entity Extraction
• Information Extraction

Summary

Named Entity Recognition (NER) is an essential NLP technique that identifies and classifies important entities in text such as people, organizations, locations, and dates. It converts unstructured text into structured, meaningful data.

NER plays a crucial role in search engines, chatbots, healthcare systems, and information extraction applications.

Conclusion

NER is a powerful tool in Natural Language Processing that helps machines understand real-world information from text. It is widely used in modern AI systems and continues to improve with deep learning and transformer models.