Combining Textual Content and Non-Textual Features of Digitized Newspaper Advertisements to Study Historical Developments in the Dutch Consumer Society

Melvin Wevers was a researcher-in-residence at the KB in 2017.

His project ‘Combining Textual Content and Non-Textual Features of Digitized Newspaper Advertisements to Study Historical Developments in the Dutch Consumer Society’ explored the question: How can textual analysis be used to study the role of the United States as a reference culture in 20th-century Dutch newspaper debates about consumer goods?

Project

Melvin Wevers explored the domain of visual advertisements and changing visual styles in this domain. His blogpost discusses the challenges of collecting and working with this kind of data, which tends to be hard to query for. 

The project resulted in the SIAMESE tool, which uses a Convolution Neural Network (CNN) to suggest visual trends by presenting users with visually similar images to a given source image. The underlying dataset SIAMESET, consisting of images and metadata of advertisements between 1945 and 1995, is also available.