Geon Diagrams: a Perception Based Method for Visualizing Structured Information

Download Publication File

Abstract

Abstract information is commonly structured, presented and communicated using diagrams. One particular type of diagram that has played a predominant role in the Information Sciences is the node-link diagram. Nodes that represent entities, objects, or units and links that represent relationships of various kinds between the nodes characterize this type of diagram. Visually these nodes are represented using outline forms such as boxes and circles and the links have been depicted with lines consisting of different characteristics.

Recent advances in perception present evidence that our visual system is endowed with the capability of recognizing objects from their structural composition. This structure is inherently 3D in nature. This thesis investigates methods for applying these theories to creating diagrams that are more expressive tools for communicating abstract information. A set of rules for creating diagrams is abstracted from vision research and collectively these define a new kind of diagramming convention called a Geon Diagram.

A series of experiments was carried out to evaluate the effectiveness of Geon Diagrams. The results indicate that applying theories of structural object recognition to the construction of node-link diagrams can facilitate identification, recall, and understanding of diagram structures. The benefits are significant in comparison to diagrams created with box-and-line notations, or diagrams with elements consisting of 2D solid outline shapes. The results show that a diagrammatic notation based on perceptual semantics, such as those available from structural object recognition theories, can facilitate intuitive comprehension of abstract problem descriptions by both novice and expert users in a given problem domain.

Citation

Pourang Irani. 2002. Geon Diagrams: a Perception Based Method for Visualizing Structured Information. PhD thesis, University of New Brunswick. (Nominated for Best Dissertation Award)

Bibtext Entry

@PHDTHESIS { PourangIraniPhDThesis,
    AUTHOR = { Pourang Irani },
    TITLE = { Geon Diagrams: a Perception Based Method for Visualizing Structured Information },
    SCHOOL = { University of New Brunswick },
    ADDRESS = { Fredricton, New Brunswick, Canada },
    YEAR = { 2002 },
    NOTE = { Nominated for Best Dissertation Award },
}