BUY WEBSITES
BUY DOMAINS
LOGIN
SIGNUP
Entity Extractor
An entity extractor scans a block of text and surfaces the names that matter, the people, brands, places, products, and acronyms a search engine would treat as the topical entities of the page. Knowing which entities your copy actually mentions, and how often, helps you check that a draft covers the people and organizations a query expects, spot thin coverage of a key brand, and compare your text against a competitor’s on the same subject.
This tool is free and runs entirely in your browser. Paste text, press Extract, and it returns a frequency-ranked table of likely entities using a transparent capitalisation heuristic. Nothing is uploaded, so it is safe to run on unpublished drafts, client copy, or confidential briefs.
🔒 Private: everything runs in your browser. Nothing you paste is uploaded.
How the entity extractor works
The tool reads your pasted text, finds capitalised words, consecutive Title-Case phrases, and ALL-CAPS acronyms, drops common sentence-start filler, dedupes case-insensitively, and ranks what remains by how often it appears.
It is a capitalisation heuristic, not a trained NER model
True named-entity recognition uses a machine-learning model trained on labelled text to decide whether a word is a person, place, or organization. This tool does something simpler and fully transparent: it treats capitalisation as the signal. Capitalised words, runs of consecutive Title-Case words, and short all-caps acronyms become candidates. That keeps it fast, private, and free of any model or server, but it means the list is candidates to review rather than a guaranteed classification. Lower-case entities slip through, and an occasional capitalised non-entity sneaks in, which is why the output is framed as likely entities.
Sentence-start words are filtered to cut false positives
Every sentence begins with a capital letter, so the first word is capitalised whether or not it is an entity. Words like The, However, and Meanwhile would otherwise flood the results. The extractor splits the text into sentences and drops a capitalised first word when it matches a built-in list of common function words and connectors. A genuine entity that happens to open a sentence, such as Google in Google shipped the update, is still kept, because it is not on the filler list. This single rule removes the most common source of noise in a capitalisation approach.
Phrases, acronyms, and counts give you a usable ranking
Consecutive capitalised words are grouped into a single phrase, so European Union and San Francisco appear as one entity each rather than four loose words, while a connector like and correctly splits Microsoft and Apple back into two. All-caps tokens of two to six letters are tagged as acronyms and never absorbed into a neighbouring phrase, keeping NASA and BBC distinct. Everything is deduped case-insensitively and counted, then ranked by frequency so the entities your text leans on hardest sit at the top, ready to copy as a tab-separated list.
Entity extractor frequently asked questions
Q1Is this a real NER (named-entity recognition) model?
No, and the tool is honest about that. It uses a capitalisation heuristic, flagging capitalised words, Title-Case phrases, and short acronyms, rather than a trained machine-learning model. That makes it instant, private, and free, but the output is a list of likely entities to review, not a guaranteed classification. It can miss lower-case entities and occasionally flag a capitalised word that is not an entity.
Q2Does my text get uploaded or stored anywhere?
No. The entire extraction runs in your browser with client-side JavaScript. Nothing you paste is sent to a server, logged, or saved, and the tool keeps working offline once the page has loaded. That makes it safe for unpublished drafts, client work, or any confidential text you would not want to send to a third-party service.
Q3Why is a capitalised word missing or an extra one included?
The tool relies on capitalisation, so a lower-case entity, like a brand styled in lower case, will be missed, and a capitalised common word can occasionally appear. Sentence-start filler words are filtered, but the heuristic is not perfect. Treat the table as a fast first pass and review it, rather than as a final, authoritative entity list for the text.
Q4What counts as a phrase versus a single entity?
Consecutive capitalised words are merged into one phrase, so San Francisco or Amazon Web Services is a single entry. A lower-case connector such as and breaks the run, which is why Microsoft and Apple becomes two separate entities. A small set of name connectors like of is allowed inside a phrase, so Bank of America stays whole. All-caps acronyms are always kept separate.
Q5How can I use the extracted entities for SEO?
Use the ranked list to check topical coverage. Compare the entities in your draft against those a competitor uses on the same query, confirm the people, brands, and places a searcher expects are actually present, and spot a key entity that appears only once when it should anchor the page. The frequency counts show which names your copy emphasises and which it barely mentions.
