Exploring the Use and Evaluation of Glyphs in Visual Analytics Applications.

Abstract

Glyph-based visualizations are widely used in visual analytics (VA) systems, where data glyphs act as data-driven visual entities that use multiple visual channels to encode multivariate data attributes. In modern VA systems, many glyphs are no longer based on a single design form, such as a single circular or radial design. We refer to those designs as composite glyphs, where a glyph is divided into multiple parts and each part uses a different visual design to encode a subset of the data. While prior work has studied glyph design principles and visual encodings, less is known about how glyphs are designed and evaluated in application-driven VA systems. In this paper, we present a systematic review of glyph-based visualizations in VA systems over the past decade. We analyze 69 papers published between 2015 and 2024 in major visualization and HCI venues, including IEEE VIS, IEEE TVCG, EuroVis, and CHI, covering 86 distinct glyph designs across diverse application domains. We analyze these glyphs along two aspects. The first is the design space, including glyph design types, combination patterns, chart contexts, supported tasks, and application scenarios. The second is verification and evaluation, including the use of design alternatives, case studies, and expert interviews or user studies. Our analysis reveals recurring patterns in how glyphs are constructed, combined, and evaluated in practice. Based on these observations, we identify five gaps in current practice and outline corresponding directions for future research.