Extractive Text Summarizer
Summarise long text by picking out its most significant sentences using Luhn's classic algorithm. Statistical only, and it runs in your browser.
This is a statistical extractive summariser. It scores and selects sentences from your own text using Luhn’s 1958 algorithm — no AI is involved, nothing is generated or rewritten, and no text is sent anywhere.
What is the Extractive Text Summarizer?
Extractive Text Summarizer shortens a long piece of writing by selecting the sentences that carry the most weight, then returning them in their original order.
- Luhn's 1958 sentence-scoring algorithm, implemented in full
- Sentences are copied verbatim — nothing is generated
- Summary length by sentence count or percentage
- Adjustable keyword cut-off and cluster gap
- Shows the terms the ranking treated as significant
- No AI, no network requests, no data leaves your browser
How to use the Extractive Text Summarizer
- 1
Paste the text you want to shorten into the input box.
- 2
Choose a summary length: a fixed number of sentences, or a percentage of the original.
- 3
Adjust the significant word cut-off and maximum key-word gap if the selection looks off.
- 4
Tick Show sentence scores to see why each sentence was chosen.
- 5
Copy or download the summary.
About the Extractive Text Summarizer
Extractive Text Summarizer shortens a long piece of writing by selecting the sentences that carry the most weight, then returning them in their original order. It uses Luhn's 1958 algorithm: common stopwords are discarded, the remaining terms are counted, and each sentence is scored by how tightly its frequent terms cluster together.
This is extraction, not generation. Every sentence in the summary is copied verbatim from what you pasted — nothing is invented, paraphrased or rewritten, and there is no AI model and no network request anywhere in the process. That makes the output predictable and easy to check against the source.
Because it runs entirely in your browser, reports, transcripts and draft articles never leave your device. You control the summary length, how selective the keyword cut-off is, and how far apart key words can sit before a cluster is broken.
Frequently asked questions
Does this summarizer use AI?
No. It is a purely statistical algorithm from 1958 that counts word frequencies and picks existing sentences. Nothing is generated or rewritten, and no text is sent to any server or model.
How does extractive summarization work?
Common words are removed, the remaining terms are counted, and the most frequent become significant. Sentences packing several significant words close together score highest, and the top few are returned in their original order.
How long should the summary be?
Around 20 to 30 percent of the original is a good starting point for articles and reports. Shorter texts usually need a fixed sentence count instead, since a percentage of six sentences is not very meaningful.
Why did it pick a sentence that seems unimportant?
The ranking is frequency-based, so a sentence repeating the document's most common terms scores well even if it reads as filler. Lowering the significant word cut-off makes the selection more selective.
Does it handle abbreviations like Dr. and U.S.?
Yes, common abbreviations, initials and dotted acronyms are recognised so they do not split a sentence in two. Very unusual punctuation can still occasionally merge two sentences into one.
Related tools
Keyword Extractor (RAKE)
Pull the key phrases out of any text with the RAKE algorithm. Ranks multi-word phrases by co-occurrence score, entirely in your browser.
Word Counter
Count words, characters, sentences, paragraphs and estimated reading time instantly. Free online word counter for essays, blogs and social media.
Character Counter
Count characters with and without spaces in real time and check limits for Twitter/X, Instagram captions, meta descriptions and SMS messages.
Paragraph Sorter
Sort blocks of text separated by blank lines alphabetically, by length, by word count or randomly, keeping each paragraph's line breaks intact.
Filter Lines (Grep for Text)
Keep or delete lines matching plain text, a wildcard or a regular expression. Includes invert match, whole word and extract-only-the-match modes.