Computer Science
Computer Science
bullet Scholarship keeps writing student’s memory, deeds alive | More

bullet Rowan University Study Links BPA Exposure to Autism Spectrum Disorder | More

bullet MEDacademy Open House Scheduled for March 14 | More

bullet Rowan University takes the handlebars in making a difference: Fourth annual Mustache March to benefit American Stroke Association | More

bullet Rowan Innovation Venture Fund accepting applications for financing | More

Computer Science Seminar Series

Using general trees insight to efficiently visualize large networks

By Chu Yao (MS Student)

Date: Wednesday April 2, 2008
Time: 11:00- 12:00
Place: Robinson 101A

Abstract:

Information visualization produces (interactive) visual representations of abstract data to reinforce human cognition and perception, thus enabling the viewer to gain knowledge about the internal structure of the data and causal relationships in it. Information hierarchies are typically modeled by an abstract tree, where nodes are entities and edges represent relationships between entities.
Because of their simpler structure, most of the research has been devoted to visualizing binary trees, which have also been classified in various ways based on their structure. However, general trees appear more commonly in practice. We therefore present our classification of general trees based on their structural characteristics.

General trees are usually drawn using planar straight-line drawings which provide an easily understandable structure to the viewer. Current general tree drawing algorithms have one common problem: their drawings on high degree trees produce many small angles, which makes it difficult to distinguish edges. The algorithm we propose allows the user to provide an angular coefficient and then employs the 'best-effort-delivery' to draw edges such that the angles are above the angular coefficient. It allows the non-root nodes to place their children within a maximum of three quads of the Cartesian plane. When a node has too many children, resulting in an impossibility of achieving angles above the specified angular coefficient, our algorithm distributes all remaining children evenly among the three quads.