When you consider the rich visual records of our past—from paintings and photographs to intricate drawings—you realise how much these archives can tell us about history and society. Yet, delving into such vast and varied collections isn’t always easy.
Doctoral researcher Tillmann Ohm at Tallinn University’s School of Digital Technologies is tackling this challenge with a fresh approach. His work shifts the focus from what an image literally represents to what it resembles. Using smart algorithms, Ohm maps images according to visual similarity, creating ‘similarity spaces’ where the closeness between points reflects how alike the images are.
At the heart of his research lies the Collection Space Navigator—a browser-based tool that lets you explore these visual connections with ease. This interactive interface helps reveal clusters and subtle patterns that traditional metadata might miss. For example, when applied to more than 200,000 frames from Soviet-era newsreels, the Navigator uncovered visual motifs and shifts in propaganda that could easily escape a manual review.
Defended on 11 June, Ohm’s research marks an important step forward in cultural data analysis. By questioning standard methods and opening up new perspectives on visual culture, his work not only deepens our understanding of cultural narratives but also encourages collaborative research across disciplines.
