Vogue was used by students and librarians at Yale to answer questions regarding women’s studies and how the answers to said questions have evolved over time. With 2,700 covers and 400,000 pages of columns and journalism completed by Vogue authors from over a century ago to the mid 2010s, both qualitative data regarding the magazine’s contents and quantitative data such as circulation statistics were applied to various mini projects.
What data was extracted from the magazines?
Specifically, In terms of the articles, the words and topics used within them were used for content modeling. Some categories of words include ‘society’, ‘food’, ‘politics’, ‘women’s health’ and, of course, fashion related terms such as ‘dress’, ‘dressmaking’, and ‘beauty’. Upon further analysis of these topics, this digital project reveals how women’s ideas surrounding these topics evolved over time.
But, how exactly was this revealed?
When dealing with unlabeled text, machine learning and numerical optimization can be used to group together related words and create a related distribution to represent the given text. This is referred to as topic modeling and was done as one of the many Vogue related projects made at Yale. These distributions were portrayed in a word cloud where users can see the frequency of certain words and phrases mentioned over the years. Extending upon this, categorizing these words over a timeline, allows for conclusions to be derived by noticing the evolution of women’s opinions and thoughts regarding the categories about life previously mentioned.
How was this timeline represented?
Through plotting saturation of categories over time, line graphs were created for each of the 50 modeled topics. Users are also able to scroll over points on the timeline and have access to the related articles from that specific point in time. The interface and interactive side of this project is extremely intuitive and allows for users to easily explore what women of the past thought about topics we now consider daily. In the example below, it’s clear that ‘politics’ was discussed to a minimum during the early editions of Vogue and started to gain prevalence at peak years like the mid 40s and 80s. While this is one of my takeaways from this project, I question what succinct findings the creators concluded based on this work.
1 thought on “Reverse Engineering: More Than A Magazine”
Comments are closed.
This is an interesting project that utilizes magazine data for women’s studies. I like your approach of deconstructing the project by asking questions, which makes the blog post very engaging. The introduction is also clear and even if I am not clicking through the website I could get an idea of the project. Your questions of what succinct findings can be concluded brings me another question: should digital projects always make a point?