word-cloud — Lattice component

word-cloud

EvidenceCanvasSeries

Spiral-packed word cloud — items sized by weight.

Open in Playground
Variant

Use for qualitative summaries — retrospective themes, survey verbatims. Word size encodes frequency or weight; not a precise data viz.

tag-cloudthemesproportion

When to use

  • Qualitative themes at a glance. Retrospective summaries, survey verbatims, theme extraction, sentiment scans. The cloud lands what the corpus is about; the silhouette and the biggest words are the read.
  • Weight is approximate, not exact. Word size encodes relative frequency or weight, but the eye reads 'biggest' and 'second biggest' before any precise ratio. Reach for word-cloud when the rank matters more than the count. A 'size = frequency' key sits in the right rail to make the encoding explicit.
  • Eight to twenty items. Below eight the spiral looks bare and the layout wastes the canvas. Past about twenty the smallest words become unreadable. Trim the long tail or cap the cloud at a 'top 15' before authoring.

When not to use

  • Precise comparisons. If the audience needs to know that 'manifest' is 1.6× 'function', the spiral packing actively misleads. Use progress or a bar chart where the eye can compare lengths directly.
  • Two or three words. A three-word cloud is a list with extra steps. Use stats for a metric row or big-number for a single weighted headline.
  • Multi-word phrases. Each li should be a single token. Multi-word phrases blow out the layout and crowd the spiral; if your data is phrases, normalise to keywords first or use quote for verbatim text.

Common mistakes

  • Assuming a zero, negative, or non-numeric weight breaks the sizing scale. None of these break anything — weight is min-max normalized across the whole cloud, so zero and negative numbers still get a well-defined proportional size, and a non-numeric or non-finite weight (e.g. a typo) is silently mapped to the MIDDLE of the scale rather than erroring. That silence is the actual hazard: a mistyped weight doesn't fail loudly, it just quietly lands a word at a plausible mid-size.
  • Using ascending rank numbers (1st place, 2nd place, …) as the weight, expecting rank 1 to render biggest. Weight is a MAGNITUDE where higher means bigger — a rank-style scale where '1' means 'most important' renders as the SMALLEST word, not the biggest. Invert rank into magnitude before authoring (the top-ranked term gets the highest number).

Authoring

<!-- _class: word-cloud -->

## What the team called out this quarter.

- velocity `12`
- ownership `9`
- handoffs `7`
- review `5`

Slots

SlotSelectorRequiredDescription
titleh2yesSlide heading framing the cloud.
wordsul > liyesOne li per word. Format: word weight`` where weight is any positive number — a frequency count, a 1–5 rating, a percentage. Words are sized and colored RELATIVE to each other: the lightest maps to small/muted, the heaviest to the hero size/accent.

Anatomy

┌─────────────────────────────────────────┐
│  header                                 │
│         Weighted words heading          │
│                                         │
│         alpha   BIG_TERM   beta         │
│          emergent   HUGE                │
│            minor   medium               │
│         tiny      LARGE   keyword       │
│  footer                           1/19  │
└─────────────────────────────────────────┘

Variant decision rule

  • default (no modifier)The standard spiral pack — the plainest read for a moderate-sized cloud.
  • constellationThe words should scatter like a starfield rather than pack tightly — a lighter, more open composition.
  • denseThe corpus is large (15-20 items) and needs to pack tightly to fit without shrinking below legibility.
  • spectrumWeight should also read as a color gradient, not just size — reinforces rank with a second visual channel.
  • focalOne single term dominates the theme and deserves an outsized size ceiling — every variant already centers the top-weight word at the spiral's origin; focal widens the size range so the leader dwarfs the rest, rather than changing where it sits.

Variants

constellationconstellation

Words scattered like stars.

<!-- _class: word-cloud constellation -->

## constellation scatters the words like stars.

- component `5`
- manifest `4`
- function `3`
- form `3`
- substance `2`
- gallery `1`

densedense

The cloud packed tight.

<!-- _class: word-cloud dense -->

## dense packs the cloud tight.

- component `5`
- manifest `5`
- function `4`
- form `4`
- substance `4`
- gallery `3`
- folder `3`
- variant `3`
- universal `2`
- cascade `2`
- scaffolder `2`
- bundler `1`
- transform `1`
- selector `1`
- palette `1`

spectrumspectrum

Words colored along a scale.

<!-- _class: word-cloud spectrum -->

## spectrum colors the words along a scale.

- component `5`
- manifest `4`
- function `4`
- form `3`
- substance `3`
- gallery `2`
- variant `2`
- universal `1`

focalfocal

One word crowned the center.

<!-- _class: word-cloud focal -->

## focal crowns one word the center.

- variants `5`
- gallery `2`
- manifest `2`
- docs `1`
- declared `1`

Related

  • progressthe weights need precise visual comparison
  • statsthe headline metrics are independent numbers, not a corpus
  • piechartthe items are parts of a whole, not free-form themes
  • quotethe verbatim language matters more than the frequency
  • listsingle-line takeaways — the takeaway variant