Named Entity Recognition (NER)
Last updated June 11, 2026
What is Named Entity Recognition in simple terms?
In simple terms, named entity recognition is a highlighter that picks out the names in text — people, companies, places, dates — and labels each one, turning plain writing into a tidy list of who, what, and where.
What is Named Entity Recognition?
Named entity recognition is an AI technique that scans text and automatically identifies and labels the named things in it — such as people, organizations, places, dates, and amounts — pulling structured information out of unstructured writing.
Named entity recognition (NER) is the AI task of finding the specific named things inside a piece of text and tagging what each one is. Reading a sentence, it picks out the people, organizations, locations, dates, money amounts, and other concrete entities, and labels them by type. The goal is to convert free-flowing writing into structured information — to take a paragraph a human wrote and pull from it a clean list of the names and facts it contains, ready for a computer to use.
The challenge is that the same word can be different things depending on context, and names don't come conveniently flagged. 'Jordan' might be a person, a country, or a brand; 'Apple' could be the fruit or the company; 'May' might be a month, a name, or a verb. NER has to use the surrounding words to decide which meaning applies and where each entity begins and ends. It also has to catch entities it has never seen before — a brand-new company name, an unfamiliar place — by recognizing the shape and context of a name rather than memorizing a fixed list. Modern systems built on language models do this well because they read each name in the full context of its sentence.
Named entity recognition is a foundational tool in natural language processing, often the first step that makes other work possible: once you know what a document is talking about, you can search it, organize it, or connect it to other information. It frequently teams up with sentiment analysis — identifying not just that an opinion is negative but exactly which product or company it targets — and feeds knowledge bases and search systems. It powers tools that scan news, process documents, organize medical or legal records, and pull key facts out of huge volumes of text far faster than any person could.
Real-world example of Named Entity Recognition
A media-monitoring service reads tens of thousands of news articles an hour on behalf of its clients. For one client — a large company — what matters is every mention of their executives, their rivals, and the places they operate. Named entity recognition is what makes this possible at that scale: as each article streams in, the system automatically pulls out and labels every person, organization, and location named in it. So the instant an article anywhere quotes the client's CEO by name, the system has already tagged "[CEO name] — Person" and "[Company] — Organization," and the mention lands in the client's morning briefing. No one is reading every article looking for those names; the highlighter does it automatically across the whole flood of text, turning a firehose of prose into a sorted list of who was mentioned where.
Related terms
Frequently asked questions about Named Entity Recognition
What is the difference between named entity recognition and sentiment analysis?
They extract different things from text. Named entity recognition pulls out the named items — who, what company, which place, what date. Sentiment analysis judges the emotional tone — positive, negative, or neutral. One identifies the subjects in the text; the other measures the feeling about them. They're often used together: named entity recognition finds that a post mentions a particular brand, and sentiment analysis adds that the post is angry — so a system learns not just that someone is unhappy, but exactly what they're unhappy about.
How does named entity recognition work?
It scans text and labels spans of words as entities of particular types — person, organization, location, date, amount, and so on — using the surrounding context to decide. The context is essential because the same word can be different things: 'Jordan' could be a person or a country depending on the sentence. Modern systems use language models trained on large amounts of text already marked up with correct entity labels, which lets them recognize even unfamiliar names by their shape and context rather than relying on a fixed list of known names.
What is named entity recognition used for?
It's used to turn unstructured text into structured, usable information. News and media monitoring uses it to track mentions of people and companies; search engines use it to understand what a query or document is about; businesses use it to extract key details from contracts, invoices, and forms; and healthcare and legal systems use it to organize records by the names, dates, and terms inside them. It's often the foundational first step that makes searching, sorting, and connecting large volumes of text possible.