Authors

Yuanzhe Jin

Abstract

Visualizing hierarchical text data requires balancing structural representation with semantic emphasis, which remains a challenging task. Traditional word clouds fail to convey hierarchical relationships, while treemaps, though efficient in spatial partitioning, are limited in expressing semantic features such as word importance. To address this issue, we propose WordTree, a novel visual design that combines word clouds with treemap layouts to represent hierarchical text data. By embedding word frequency visualization within the spatial structure of a treemap, WordTree effectively communicates both multi-level hierarchical structures and the relative importance of words within the dataset. Compared with traditional word clouds or treemaps, WordTree offers improved space efficiency, supports hierarchical exploration, and enables direct visual comparison across categories or subgroups. We demonstrate the effectiveness of WordTree through several case studies involving textual datasets such as news corpora and topic hierarchies. User feedback suggests that WordTree enhances understanding of data structure while maintaining a strong visual appeal. Download paper here