Posterity: Balancing historical context and visual dynamism while visualizing a collection of American labor posters

Keywords

digital humanities, image visualization, image interpretation, historical records, similarity retrieval

Abstract

Visual archives of political movements are rich cultural resources, yet often difficult to explore at scale due to complex visual semantics and limited interaction models. We present Posterity, an interactive visualization system for 784 digitized American labor posters (1900–2010), designed to support both historical contextualization and visual-semantic exploration. Posterity integrates curated metadata, CLIP-based multimodal embeddings, and unsupervised clustering to offer three coordinated views: a timeline aligned with key labor events, a 3D semantic cloud, and a similarity spiral responsive to image-, object-, or gesture-based input. Together, these views enable users to trace recurring visual motifs, discover rhetorical patterns, and explore labor movement narratives from multiple entry points. While developed for labor posters, the approach demonstrates potential for adaptation to other visual cultural heritage collections, particularly those with rich metadata and symbolic content.