scattertext
Create interactive HTML scatter plots visualizing how language differs between document types.
Generate interactive HTML scatter plots to find distinguishing terms in text corpora.
Scattertext is a Python library that builds browser-based interactive visualizations of how language differs across document categories. It helps you analyze term frequencies, associations, and semantic patterns using non-overlapping labels. Data scientists and text analysts exploring textual datasets will find it especially useful.
What it is
It takes parsed document corpora and produces interactive D3 scatter plots where points correspond to terms or phrases positioned by category usage. Features include label placement algorithms to prevent overlap, zoom and pan controls, and support for alternative scoring methods like Cohen's d, Hedge's g, and Bi-Normal Separation. You can also visualize topic models, word embeddings, and specialized dictionaries like Empath or Moral Foundations.
Who it's for
It suits Python developers and computational social scientists needing to explain classifier weights or compare text corpora. It does not suit users who do not know Python or those looking for a drag-and-drop web dashboard without coding.
Key features
Interactive D3 Scatter Plots
Build browser-based visualizations with non-overlapping term labels and zoom support.
Multiple Scoring Methods
Score terms using scaled F-score, Cohen's d, Hedge's g, Cliff's Delta, or Bi-Normal Separation.
Specialized Dictionary Support
Visualize Empath topics, Moral Foundations 2.0 categories, and SentencePiece tokens.
Model Interpretation
Explain scikit-learn text classification weights and topic models through interactive projections.
Up and running in 3 steps
Our honest verdict
What's great
- Generates standalone interactive HTML files for easy sharing
- Smart label positioning prevents text overlap on dense plots
- Supports multiple statistical effect size metrics out of the box
- Integrates with popular NLP libraries like spaCy, gensim, and jieba
Where it falls short
- Requires proficiency in Python programming and text preprocessing
- Heavy computations can be resource-intensive for very large corpora
- Best performance requires installing multiple external NLP dependencies
Pricing & license
Free under the Apache-2.0 license. There are no paid hosted options provided by the project, so the only cost is running it yourself in your Python environment.
Open remote AI-training roles
870 open remote AI-training roles, updated November 14, 2026; most pay $20–$170 an hour. Zekai lists them free; the application goes to the employer’s platform.
Alternatives
Use it, or have us set it up
FAQ
Can I use Scattertext without installing spaCy?
Yes, you can substitute spaCy with scattertext.WhitespaceNLP.whitespace_nlp, though tokenization and sentence boundary detection will rely on basic regular expressions.
What browsers are best for viewing the output?
The generated HTML outputs look and perform best in Chrome and Safari.
How do I prevent labels from overlapping?
Scattertext automatically applies a layout algorithm that selectively labels points so they do not overlap with other labels or points.
