Ninth International Conference on Communication & Media Studies

Abstract

We created a web-based prototype of a digital visualization tool, Faces in Time, that allows users to investigate images of faces extracted from the pages of Time magazine. Our archive contains 3,389 issues ranging from 1923 to 2014, and the data comprises 327,322 faces extracted from the archive, each categorized by gender, as well as detailed characteristics (including facial expression) of a subset of 8,789 of those faces. Faces in Time (URL: https://magazineproject.org/sandbox/ ) contains 5 interactive visualizations that invite visitors to explore our data and to critically examine how images of human faces fit within broader cultural and historical contexts. This work applied machine learning techniques in its various facets, from data collection, to classification, to visualization. I discuss here the limits of using machine learning techniques in the humanities, as well as the opportunities it provides for exploring large data sets and for providing alternative (non-human) perspectives on the data. The website uses contemporary techniques to defamiliarize images of faces, taking them out of the magazine narrative context, and instead providing bird’s eye views of the cultural and historical context in which the faces were found. While defamiliarization is primarily an artistic technique to (re)open a viewer’s eyes to the real, it is also productive in digital history. Digital tools afford readers the ability to interactively re-contextualize images, and the hope here is to design platforms that can encourage critical thinking, and that position website visitors as investigators rather than passive consumers of visual and media culture.