📝 Text Summarizer

Pull out the most representative sentences from a passage using simple word-frequency scoring — a classic "extractive" summarizer. It picks existing sentences rather than generating new text.

How a sentence earns its place

Every word is counted first, ignoring about seventy stopwords (the, is, with) and anything under three letters. Each sentence is then scored:

            sum of the counts of its content words
score  =  ------------------------------------------
              number of words in the sentence

Repetition is the whole signal: a word used four times lends weight to all four sentences it appears in. The highest scorers are put back in their original order and joined, so the summary still reads forwards. Nothing is rewritten - every sentence is yours, verbatim.

The divisor matters. In the sample text longer appears twice, enough to score the six-word opening sentence a perfect 1.00 and put it above the 0.75 of the long sentence that explains what the tool does.

When it works, and when it does not

Extraction can invert meaning. Lift a strong claim out and leave the finding was later withdrawn behind, and the summary now asserts something the original retracted. Read it against the source before passing it on.

Frequently asked questions

Does this use AI to write the summary?

No. It is a word-frequency ranking that runs in the page: nothing is uploaded, no model is called, and it never invents a sentence.

How much text does it need?

A few paragraphs. If the passage holds no more sentences than the number you asked to keep, it returns all of them and says so, rather than shortening any of them.

Why did it choose the shortest sentence?

Because the score is an average, not a total. A brief sentence made almost entirely of repeated topic words beats a long one that carries the same words plus a lot of filler.