QR コード生成

What a word cloud shows — and what it hides

A word cloud turns a block of text into a picture where the most frequent words are the biggest. It is a fast way to see what a document is about without reading it: the dominant themes leap out as the largest words. This generator does that entirely in your browser, drawing onto an HTML5 canvas, so the text you paste never leaves your device and there is no upload step. Used well it is a genuine summarising tool; used carelessly it can mislead, and knowing the difference is the whole skill.

How the frequency count is built

Before anything is drawn, the tool counts words, and the counting rules are what make the result meaningful. The text is lowercased and stripped of punctuation, then split into words. Three filters then apply:

• Stop words are removed. Around 150 extremely common function words — the, and, of, is, to, a — are discarded. These dominate any English text but carry almost no meaning, so leaving them in would produce a cloud that just says "THE AND OF" in giant letters. Filtering them is what lets the real content words surface.
• Very short words are dropped. Anything under three characters is excluded, removing leftover fragments and stray initials.
• Case and surrounding marks are normalised. "Data," "data" and "DATA" all count as the same word, so a term is not split across three smaller entries.

What remains is tallied, sorted from most to least frequent, and truncated to the limit you set with the Max Words slider (10 to 100). Only those top words are drawn.

How the layout is drawn

Word size is proportional to frequency. The most common word is rendered at the largest size, the rarest survivor at the smallest, and everything else scales linearly in between — so a word drawn twice as large appeared roughly twice as often, give or take the rounding. Words are then placed using a spiral algorithm: the biggest word goes near the centre, and each subsequent word spirals outward from the middle, testing positions until it finds one that does not overlap any word already placed. This is why a word cloud is denser in the centre and airier at the edges, and why bigger words tend to anchor the middle.

A worked example

Paste a few paragraphs about machine learning. The counter throws away every the and of, then discovers that data appears nine times, learning seven, model five, and training four. With Max Words at 60, all of these make the cut. data, the most frequent, is drawn largest and lands at the centre; learning and model appear noticeably smaller and spiral out around it; one-off mentions like healthcare sit small near the edge. Hover over any word and a tooltip reports its exact count — for example "data": 9× — so you are never guessing at the numbers behind the sizes. When the picture looks right, choose a colour scheme (Random, blue-to-purple Gradient, a Single Color, or the Warm palette) and press Download PNG to save it. The canvas renders at your display's pixel density, so the saved image is sharp on high-resolution screens.

Real use cases

• Survey and review analysis. Paste hundreds of open-ended responses to instantly see which words customers reach for most — a quick read on sentiment and recurring complaints.
• Summarising long documents. Drop in an article, report, or transcript to preview its main topics before deciding whether to read it in full.
• Teaching and presentations. Generate a cloud from a speech or a chapter as a discussion prompt or a slide visual.
• Editing your own writing. See which words you lean on too heavily; an unexpectedly huge word is often a crutch worth varying.

Tips for a more honest cloud

• Feed it enough text. A few sentences produce a cloud where almost every word appears once and sizing is meaningless. Frequency only tells a story when words actually repeat, so paste a substantial sample.
• Tune Max Words to the goal. A low limit (around 20) gives a clean headline of the dominant themes; a high limit (80–100) shows the long tail but can look cluttered.
• Mind that it counts forms separately. "Run", "running", and "ran" are three different words to a frequency counter. If a concept is split across several forms, its true importance may be understated.
• Read it as a frequency chart, not a meaning chart. A word cloud only knows how often words appear, never how they are used. Sarcasm, negation ("not good"), and context are invisible to it — a big word tells you a topic was discussed, not what was said about it.

Privacy note

The text you paste is processed and rendered entirely in your browser on a local canvas. Nothing is uploaded, and the PNG you download is generated on your device. You can use it with your network disconnected and confirm that sensitive text never leaves your machine.