Question 1

How does the paragraph summarizer work?

Accepted Answer

Paste your text (article, report, notes, or transcript) and the tool splits it into paragraphs and sentences, then scores every sentence using three signals: TF-IDF (rare, salient words score higher), position (first and last sentences of each paragraph get a bonus), and keyword presence (sentences containing the document's top keywords score higher). The top-scoring sentences are reassembled in original order to form the summary. This is extractive summarization — no model download, no network, no upload.

Question 2

What length options are available?

Accepted Answer

Three presets: Short (~15% of sentences), Medium (~30%), and Long (~50%). You can also switch between paragraph output (a single flowing summary) and bullet output (one bullet per selected sentence). For very long inputs (over ~4,000 words) the tool automatically runs map-reduce chunking: each chunk is summarized, then the chunk summaries are combined into a final summary.

Question 3

Can I see key points and the compression ratio?

Accepted Answer

Yes. The tool extracts the top 5–10 key points (highest-scoring sentences across the whole document), shows a compression-ratio readout (summary word count ÷ original word count), and reports sentence + word counts before and after. You can copy the summary, download it as .txt or .md, and share a URL with the input + settings encoded (the input is stored in the URL hash, not on any server).

Question 4

What extra features does this tool have compared to others?

Accepted Answer

(1) Extractive TF-IDF + position + keyword scoring. (2) Three length presets (short/medium/long). (3) Paragraph or bullet output. (4) Multi-paragraph support with per-paragraph scoring. (5) Map-reduce chunking for long docs. (6) Top key-points extraction. (7) Compression-ratio + word/sentence counts. (8) Copy + .txt + .md download. (9) Shareable URL with input + settings encoded. (10) Local history (max 20). (11) Side-by-side original vs summary. (12) Sample text button. (13) Honesty disclaimer (extractive ≠ abstractive — verify before quoting). (14) Optional BYO-key LLM hook for GPT-class abstractive quality.

Question 5

Is my text sent anywhere?

Accepted Answer

No. All paragraph splitting, sentence tokenization, TF-IDF computation, scoring, and rendering run locally in your browser. Your text never leaves this device. History is stored in localStorage on this device only. The optional BYO-key LLM path (if you supply an API key) sends text directly from your browser to the provider you choose — UnQTools never proxies or stores it.

AI Paragraph Summarizer

How to use