Problem
Research questions and friction points this paper is trying to address.
font
impression
style specificity
co-embedding
hyperbolic space
Hierarchical Co-Embedding of Font Shapes and Impression Tags
🤖 AI Summary
This work addresses the non-bijective relationship between typeface shapes and impression labels, where different impressions impose varying degrees of stylistic constraint—referred to as style specificity. The authors propose a hyperbolic co-embedding framework that, for the first time, models style specificity through radial geometric structures, capturing semantic constraints via hierarchical entailment rather than simple pairwise associations. This approach jointly represents font images and textual impression labels in a shared multimodal space, enabling nuanced modeling of both font-to-impression mappings and relationships among impressions ranging from low to high specificity. Evaluated on the MyFonts dataset, the method significantly outperforms strong one-to-one retrieval baselines. The resulting embedding space not only offers an interpretable quantification of style specificity but also naturally encodes a coherent transition from vague to concrete typographic impressions.
📝 Abstract
Font shapes can evoke a wide range of impressions, but the correspondence between fonts and impression descriptions is not one-to-one: some impressions are broadly compatible with diverse styles, whereas others strongly constrain the set of plausible fonts. We refer to this graded constraint strength as style specificity. In this paper, we propose a hyperbolic co-embedding framework that models font--impression correspondence through entailment rather than simple paired alignment. Font images and impression descriptions, represented as single tags or tag sets, are embedded in a shared hyperbolic space with two complementary entailment constraints: impression-to-font entailment and low-to-high style-specificity entailment among impressions. This formulation induces a radial structure in which low style-specificity impressions lie near the origin and high style-specificity impressions lie farther away, yielding an interpretable geometric measure of how strongly an impression constrains font style. Experiments on the MyFonts dataset demonstrate improved bidirectional retrieval over strong one-to-one baselines. In addition, traversal and tag-level analyses show that the learned space captures a coherent progression from ambiguous to more style-specific impressions and provides a meaningful, data-driven quantification of style specificity.
Innovation
Methods, ideas, or system contributions that make the work stand out.
hyperbolic embedding
style specificity
entailment modeling
font-impression correspondence
hierarchical co-embedding
🔎 Similar Papers
No similar papers found.
