Robots Reading Vogue

Data Mining is in Fashion

Few magazines can boast being continuously published for over a century, familiar and interesting to almost everyone, full of iconic pictures — and also completely digitized and marked up as both text and images. What can you do with over 2,700 covers, 400,000 pages, 6 TB of data? Students, librarians and faculty are excited about the possibilities of working with Vogue to explore questions in fields from gender studies to computer science. We highlight some early experiments below:

Word Vectors

Using temporal word embeddings to study the shifting notion of beauty in Vogue, by Sydney Bowen '21.

Explore »

Cover Averages

Visual continuity and change across the decades.

Explore »

n-gram Search

Search and compare word usage in 400,000+ pages within ads, articles, or all texts.

Explore »

Topic Modeling

Self-organizing themes determined by word coöccurrence.

Explore »

Advertisements

Sort ads by frequency, date, and industry.

Explore »

Statistics

View figures on circulation, ratio of articles to advertisements, price per issue, and number of pages per year.

Explore »

Student Work

See projects conducted by Yale students using this data.

Explore »

Colormetric Space

Hue, Saturation and Lightness.

Explore »

FabricSpace

Using Word Embedding Models (word2vec) to explore hierarchical clusters of fabric types.

Explore »

Take a Memo

Algorithmically-generated memos, in the style of Vogue editor-in-chief Diana Vreeland.

Explore »

Slice Histograms

Direct visualization of color patterns.

Explore »