Files
2026-02-10 10:06:54 -05:00

5.3 KiB
Raw Permalink Blame History

Local Word Cloud Tool (DOCX or Ranked CSV)

A single-file, local HTML tool that:

  • extracts text from a DOCX in your browser,
  • generates a ranked word list (word → count),
  • renders a word cloud, and
  • exports the result as CSV, SVG, or PNG.

No server required. Everything runs client-side in your browser.


Files

  • wordcloud_tool.html (the only file you need)

Optional inputs:

  • Any .docx file (e.g., a list of titles, abstracts, bullets, etc.)
  • A ranked .csv file with columns: word,count

How to run

  1. Save the tool as wordcloud_tool.html.
  2. Open it in a modern browser (Chrome or Edge recommended).
  3. Choose one input method:
    • DOCX → Build list from DOCX
    • CSV → Build cloud from CSV

Tip: If your organization blocks external CDNs, see Offline Mode below.


1) Load DOCX

Click Load DOCX and select any .docx file.

You can upload any DOCX file—Word documents, exported DOCX from Google Docs, etc.
The tool extracts the document’s text in-browser (no upload to a server).

2) Build the word list

Click Build list from DOCX.

The tool:

  • extracts text,
  • normalizes punctuation,
  • optionally removes years (e.g., 1987, 2024),
  • tokenizes into words,
  • applies stopwords and filters,
  • counts occurrences,
  • sorts by frequency.

3) Tune the output (important)

Use these controls:

  • Top N words
    Limits the ranked list to the top N words after filtering.
    (Typical ranges: 50–200)

  • Min word length
    Filters out short tokens (e.g., “of”, “in”).
    Try 3–4 for academic titles.

  • Remove years
    Removes tokens like 1987 / 2024 if they appear in the text.

  • Drop numbers-only tokens
    Removes pure numeric tokens (still keeps alphanumeric like SiO2, GaAs).

  • Stopwords (the most important control)
    A list of words to ignore, one per line or comma-separated.
    This includes common English stopwords plus common academic “filler” words.


Editing the word outputs (how to control what appears)

You have two main ways to edit the final words in the cloud.

Method A: Edit stopwords (fastest)

If an unwanted word is dominating the cloud:

  1. Find it in the cloud or ranked list.
  2. Add it to the Stopwords box.
  3. Rebuild (click Build list from DOCX again).

Examples of common “junk” words to add:

  • study, analysis, characterization, development, effects, properties

You can also remove words from stopwords if you do want them included.

Method B: Export CSV → edit → reload (most control)

If you need precise control (e.g., merge synonyms, force counts, remove a few terms):

  1. Build from DOCX.
  2. Click Export ranked CSV.
  3. Edit the CSV in Excel (or any spreadsheet tool).
    • You can delete rows, rename words, or change counts.
  4. Save as CSV (keep word,count columns).
  5. Load the edited CSV using Load Ranked CSV.
  6. Click Build cloud from CSV.

This gives full manual control over the final word list.


Exporting graphics

Click Export SVG to download wordcloud.svg.

  • Best for Illustrator / InDesign / PowerPoint (vector)
  • You can edit text as objects (unless you later outline it)

Note: The exporter replaces the system-ui font keyword with Helvetica in the SVG to avoid Illustrator import warnings.

Export PNG

Click Export PNG to download wordcloud.png.

  • Good for quick slide insertion
  • The PNG exporter preserves aspect ratio based on the SVG viewBox (no squishing)

If you want a specific slide size (e.g., 1920×1080 or 3840×2160), the PNG export function can be adjusted in the script.


Tips for best results

  • Start with Top N = 100–150 and Min length = 3
  • Expand stopwords aggressively to remove “academic glue” words
  • If you have chemical formulas/material names:
    • they often appear as alphanumeric tokens (e.g., GaAs, SiO2) and will be kept
  • Consider exporting CSV and merging synonyms:
    • e.g., magnetic + magnetism, diffusion + diffusive

Offline mode (no internet access)

By default, the tool loads libraries from CDNs:

  • d3
  • d3-cloud
  • mammoth (DOCX parsing)

If you need a fully offline version:

  1. Download those JS files.
  2. Place them in the same folder as wordcloud_tool.html.
  3. Replace the three <script src="https://..."> tags with local paths, e.g.:
<script src="./d3.min.js"></script>
<script src="./d3.layout.cloud.js"></script>
<script src="./mammoth.browser.min.js"></script>

Troubleshooting

Illustrator warning on import

If Illustrator complains about fonts (e.g., system-ui), export SVG again—the tool sanitizes that token to Helvetica for compatibility.

“Squished” PNG exports

PNG export uses the SVG viewBox to preserve aspect ratio. If you still see distortion, verify you’re viewing the PNG at 100% and that the viewer isn’t applying non-uniform scaling.

DOCX content looks incomplete

Some Word docs contain text inside tables/text boxes. Mammoth extracts “raw text” best from normal paragraphs. If your content is in tables/text boxes, consider copying it into a simple paragraph/list format first, or export a ranked CSV manually.