Text Summarizer

Most "summarizer" tools online do one thing: paste text in, get a fixed handful of sentences out, with zero visibility into why those particular sentences were chosen or any way to adjust how much gets kept. That's a black box, and it makes the output hard to trust for anything more serious than a quick skim.

This tool is built around TextRank, a genuine, well-established graph-ranking algorithm published by Mihalcea & Tarau in 2004. It works by treating every sentence as a node in a graph, connecting sentences that share significant words, and running an iterative ranking process — the same underlying mathematical idea as Google's original PageRank, applied to sentences instead of web pages, entirely on the client. It is an extractive summarizer: it selects and reorders your own original sentences rather than generating new text the way a language model would, which means the output is always grounded in exactly what you wrote or pasted, with no risk of a fabricated detail.

Three lenses change what gets prioritized: "Key sentences" surfaces the most central ideas across the whole text, "Numbers & stats" filters down to only the sentences that contain data — figures, percentages, dollar amounts — useful for pulling the hard facts out of a long report, and "Quotes only" isolates every directly quoted line, useful for scanning an interview transcript or article for exactly what people said. A length slider adjusts what percentage of the original sentences make the cut, regenerating instantly.

The transparency feature is the part most summarizers skip entirely: toggle "Show why these sentences were picked" and every sentence in your original text is shaded by its computed importance score, with the ones selected for the summary underlined — so instead of just trusting the output, you can see exactly how the algorithm reasoned about your text.