How to use the text summarizer
- 1Paste the text into Text to summarize (or press Example).
- 2Choose a Length: Short (15% of sentences), Medium (25%), Long (40%), or Sentences… for an exact number.
- 3Choose a Format: Paragraph, Bullet points or TL;DR (1 sentence). The summary updates as you change options.
- 4Open Which sentences were picked to see every sentence's score, then copy or download the summary.
The built-in example, Short, Bullet points
Urban trees do more than make streets look pleasant. A mature street tree shades pavement and buildings, … (4 paragraphs, 11 sentences)
• Urban trees do more than make streets look pleasant. • Planting trees is not enough on its own.
Real output: 2 of 11 sentences (15% rounded). Both open a paragraph and share the text's key term, “trees”.
How extractive summarising picks sentences
This summarizer uses the frequency method first described by H. P. Luhn at IBM in 1958. It counts how often each content word appears (common words such as “the” and “because” are ignored, and simple plurals are merged), then gives each sentence a score: the sum of its distinct words' frequencies relative to the most frequent word, divided by the square root of the number of words so long sentences don't win just by being long. The first sentence of the text gets a 20% boost and the first sentence of each paragraph 10%, because writers tend to put key points there. Sentences under 5 words score 0, and a sentence sharing 60% or more of its words with one already picked is skipped. The top sentences are shown in their original order.
Choosing summary length
Short keeps about 15% of the sentences, Medium 25% and Long 40%, rounded, with at least one. For an abstract or a meeting brief, Short is usually enough; for study notes, Medium or Long keeps more supporting detail. Sentences… sets an exact number. The footer shows how many sentences and words were kept and the share of the original. Very short inputs (fewer than 3 sentences) are left alone.
Paragraph, bullets or TL;DR
Paragraph joins the picked sentences into one block, useful for an abstract. Bullet points puts each on its own line with a bullet, which suits notes and emails. TL;DR returns only the single highest-scoring sentence, prefixed “TL;DR:”. Key terms beside the summary lists the most frequent content words, which is often the quickest way to see what a text is about.
When a summary misses context
Because sentences are lifted as they are, a picked sentence may start with “This” or “However” and refer to something that wasn't picked. Numbers and caveats from neighboring sentences can also be lost, and a text whose main point appears only once (in a conclusion, say) may be under-represented. Check the summary against Which sentences were picked, and add a sentence by hand where the meaning depends on it. For formatted web pages, strip HTML before summarising. To compare lengths, use the word counter or the reading time calculator.
Citing the original source
An extractive summary contains the author's exact words, so treat it as quotation, not your own writing. In an essay or report, either put the sentences in quotation marks with a citation, or use the summary as notes and write your own paraphrase, still citing the source. Keep the author, title, publication and date with any summary you save so you can find and cite the original later.
Is my text private?
This tool runs entirely in your browser. Your text is not uploaded to our servers.