It is often thought that fashion repeats itself. This premise is explored in the Digital Arts and Humanites Project: Robots Reading Vogue. In 2011 Yale University Library analyzed and visualized Vogue magazine’s digital archives. The analytics magazine archive produced 6 TB of data. The projects that they created with the data can be found on Yale Digital Humanities Lab website: https://dhlab.yale.edu/projects/vogue/. The goal of the project was to give students a chance to practice hone in their skills. Beyond that it’s to see how the magazine changed over time. I chose to examine the Topic modeling part of the project. This means that different word frequencies and associations were analyzed with computer algorithms.
Sources
The processed data came from the articles in Vogue magazine from 1892 to 2022.
Processes
Simply put, the project uses programmed statistical analysis to see what terms seem to be used together in articles and how popular they are each year. More technically, I suspect that all of the articles in Vogue’s archive were processed with an algorithm using Mallet framework. Mallet is a word processing Java package that can compute word clusters and frequencies.
Presentation
The presentation of the analytic work is the coolest part of the project. The final result of the project is two fold. The first are word clusters. A word cluster shows words and phrases that tend to come up together in different articles.
For example, the word art is often found around phrases like: museum art, metropolitan museum, contemporary art and words like: paintings, gallery and exhibition. The word dressmaking is commonly associated with words and phrases like: skirt, material, vogue pattern, collar cuffs and patent leather.
The second part of the project shows the frequency of each word for a year. Words like advice and etiquette were most popular around 1900 and peaked a second time in 1935. Women’s health, was very popular in the 1980s. This data is communicated through a graph that show’s its saturation and the year along with overlapping displays of Editor in Chief.
New Questions
A question that arrises for me as I examined this project is where else can this framework for archival analysis be applied? I think it would be super interesting to see the different trends in language for different word groups. What would it look like to examine the and compare the wording of laws at a local and federal level? How might the words that get used align with historical events?
