shapedescribing

Shapedescribing is a field or practice focused on conveying the form of a shape through descriptive language, enabling a listener or reader to reconstruct or understand the shape without visual cues. It spans education, accessibility, design communication, and human–computer interaction, and can be applied to both simple and complex geometries.

Core elements include the designation of the shape type (circle, triangle, polygon), quantitative attributes (size, radius, edge lengths, area, perimeter), and qualitative attributes (symmetry, orientation, curvature). Relations among parts, such as containment or adjacency, may also be described.

Methods include template-based description, where fixed phrases map to geometric properties; parameterized natural language generation that can adjust detail level; and data-driven approaches that lift descriptors from annotated corpora or inverse tasks that generate shapes from textual descriptions.

Applications include helping visually impaired users understand shapes in math education, enabling accessibility in diagrams, providing human-readable descriptions in computer-aided design, and serving as a testbed for grounding language in perception in AI research. For example, a description might read: a square with side length two units, rotated 45 degrees.

Challenges include ambiguity, variability in linguistic expression, and the need for standardized vocabularies; future directions include multimodal grounding, user-adaptive detail levels, and formal evaluation benchmarks.