Andrew Crowell

Date Approved

10-6-2010

Embargo Period

3-3-2020

Document Type

Thesis

Degree Name

M.S. Engineering

Department

Electrical and Computer Engineering

College

Henry M. Rowan College of Engineering

Advisor

Rusu, Adrian

Subject(s)

Information visualization

Disciplines

Electrical and Computer Engineering

Abstract

Information Visualization is an area of research concerned with the presentation of abstract data in a visual format. It has many diverse applications in fields of software engineering, information sciences, biology, chemistry, and medical, among others. This thesis is primarily concerned with the application of information visualization to the display of directed graphs and to the display of multivariate data for analysis. Two novel applications will be presented that are both advancements in the field of information visualization. The first application applies to the visualization and navigation of large or infinite graphs and networks. The quality of a graph drawing algorithm is often measured by its edge crossings, angular resolution, aspect ratio, and node labeling. Algorithms for drawing trees in general are segregated from algorithms for drawing graphs. This thesis will present a graph visualization system that uses a novel interconnection between a tree drawing algorithm and graph drawing techniques. First, the graph is transformed into a tree and nodes that have multiple parent connections within the graph are duplicated within the tree. While some of the connection information is lost during this transformation, the multiple connections can be regained by interactively displaying the details based on the degree of interest. The tree drawing algorithm that is used also provides a way to display a connection without drawing an edge. This edgeless visualization allows edge crossings and angular resolution issues to be eliminated. The graph is represented using a rings visualization, so it naturally has an aspect ratio of 1. Finally, a circular labeling method is used that provides user-friendly labels that do not overlap and clearly show node affiliation. The result is an interactive, navigable graph visualization system that is a useful tool for the display of large or infinite hierarchical and network graphs. The second application involves the display of multivariate, multi-set data. Generally, the purpose of graphically displaying large datasets is to find the general average of where most of the data lies and then to find the outliers (i.e. the data points that are most distant from the average). The visualization presented in this thesis will attempt to find both by using a multitude of common graphing techniques to expand upon the traditional star glyph, creating a three-dimensional visualization that allows comparisons among multiple datasets of multivariate data.

Download

Share

COinS