LICS: Towards Theory-Informed Effective Visual Abstraction of Property Graph Schemas

Property graph schemas are essential for organizing property graph data, serving both prescriptive and descriptive roles. This has led to the recent development of property graph schema languages such as PGSchema. While understanding of these languages requires familiarity with complex syntax, this poses usability challenges, particularly for domain experts who are not programmers. Current visual abstractions, such as the labeled schema graph (LSG), simplify representation but suffers from visual clutter and limited feature support. To address these challenges, we propose a novel, generic, and extensible visual abstraction, labeled iconized composite schema (LICS), whose design is informed by theories and principles from HCI, cognitive psychology, and visualization. A novel LICS-based visual interface coined PASCAL is also proposed to facilitate visualization of property graph schemas. Under the hood, it leverages the Map-Paint algorithm for creating the visual components of LICS. A user study demonstrates that LICS is superior to the traditional LSG abstraction w.r.t. usability, effectiveness, query formulation efficiency, and schema comprehension.