department of computer science software engineering, graphics, and visualization research group 15th...

17
DEPARTMENT OF COMPUTER SCIENCE SOFTWARE ENGINEERING, GRAPHICS, AND VISUALIZATION RESEARCH GROUP 15th International Conference on Information Visualisation 2011 PieVis: Interactive Graph Visualization Using a Rings-Based Tree Drawing Algorithm for Children and Crust Display for Parents Presented by: Dr. Adrian Rusu Adrian Rusu, Bryan Petzinger and Andrew Fabian Department of Computer Science, Rowan University, USA Andrew Crowell Federal Aviation Administration, William J Hughes Technical Center, USA

Post on 15-Jan-2016

228 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: DEPARTMENT OF COMPUTER SCIENCE SOFTWARE ENGINEERING, GRAPHICS, AND VISUALIZATION RESEARCH GROUP 15th International Conference on Information Visualisation

DEPARTMENT OF COMPUTER SCIENCESOFTWARE ENGINEERING, GRAPHICS, AND VISUALIZATION RESEARCH GROUP

15th International Conference on Information Visualisation 2011

PieVis: Interactive Graph Visualization Using a Rings-Based Tree Drawing

Algorithm for Children and Crust Display for Parents

Presented by: Dr. Adrian Rusu

Adrian Rusu, Bryan Petzinger and Andrew FabianDepartment of Computer Science, Rowan University, USA

Andrew CrowellFederal Aviation Administration, William J Hughes Technical Center, USA

Page 2: DEPARTMENT OF COMPUTER SCIENCE SOFTWARE ENGINEERING, GRAPHICS, AND VISUALIZATION RESEARCH GROUP 15th International Conference on Information Visualisation

Introduction

• Interactive focus+context graph visualization• Visualize large or infinite graphs• Incremental exploration

• Graph traversal through user interaction• Maintain consistent view/layout to prevent user

disorientation• Interconnects tree and graph drawing techniques

DEPARTMENT OF COMPUTER SCIENCESOFTWARE ENGINEERING, GRAPHICS, AND VISUALIZATION RESEARCH GROUP

Page 3: DEPARTMENT OF COMPUTER SCIENCE SOFTWARE ENGINEERING, GRAPHICS, AND VISUALIZATION RESEARCH GROUP 15th International Conference on Information Visualisation

WebGPS

DEPARTMENT OF COMPUTER SCIENCESOFTWARE ENGINEERING, GRAPHICS, AND VISUALIZATION RESEARCH GROUP

• Usage Example – Web visualization

• Synchronization

• As user navigates through visualization, the focal node is displayed in browser

• Complementary relationship

• PieVis shows orientation within graph and relationships between nodes

• Browser provides content

Page 4: DEPARTMENT OF COMPUTER SCIENCE SOFTWARE ENGINEERING, GRAPHICS, AND VISUALIZATION RESEARCH GROUP 15th International Conference on Information Visualisation

WebGPS

DEPARTMENT OF COMPUTER SCIENCESOFTWARE ENGINEERING, GRAPHICS, AND VISUALIZATION RESEARCH GROUP

• Usage Example – Web visualization

Page 5: DEPARTMENT OF COMPUTER SCIENCE SOFTWARE ENGINEERING, GRAPHICS, AND VISUALIZATION RESEARCH GROUP 15th International Conference on Information Visualisation

Drawing Algorithm

• Transform input graph to a tree, without losing information

• Duplicate vertices of graph

• Resulting tree has all edges, but may have more nodes than original input graph

DEPARTMENT OF COMPUTER SCIENCESOFTWARE ENGINEERING, GRAPHICS, AND VISUALIZATION RESEARCH GROUP

Page 6: DEPARTMENT OF COMPUTER SCIENCE SOFTWARE ENGINEERING, GRAPHICS, AND VISUALIZATION RESEARCH GROUP 15th International Conference on Information Visualisation

Drawing Algorithm

• (A) Directed graph• (B) Tree corresponding

to graph• (C) Connections are re-

established when more detail is shown

DEPARTMENT OF COMPUTER SCIENCESOFTWARE ENGINEERING, GRAPHICS, AND VISUALIZATION RESEARCH GROUP

Page 7: DEPARTMENT OF COMPUTER SCIENCE SOFTWARE ENGINEERING, GRAPHICS, AND VISUALIZATION RESEARCH GROUP 15th International Conference on Information Visualisation

FastRings

• Real-time visualization requires visualization to be rendered on the fly

• Original algorithm (RINGS) was bottom-up– Required entire tree to be traversed before it

could be drawn• FastRings developed from original RINGS, but as

a top-down algorithm– Rendering may begin as soon as first level of

tree is crawled

DEPARTMENT OF COMPUTER SCIENCESOFTWARE ENGINEERING, GRAPHICS, AND VISUALIZATION RESEARCH GROUP

Page 8: DEPARTMENT OF COMPUTER SCIENCE SOFTWARE ENGINEERING, GRAPHICS, AND VISUALIZATION RESEARCH GROUP 15th International Conference on Information Visualisation

Node and Subtree Highlighting

DEPARTMENT OF COMPUTER SCIENCESOFTWARE ENGINEERING, GRAPHICS, AND VISUALIZATION RESEARCH GROUP

• Conversion from graph to tree may result in missing links

• User interaction can be utilized to rediscover this information

• Hovering over a node reveals where that node occurs in other areas of the graph through highlighting

Page 9: DEPARTMENT OF COMPUTER SCIENCE SOFTWARE ENGINEERING, GRAPHICS, AND VISUALIZATION RESEARCH GROUP 15th International Conference on Information Visualisation

Node and Subtree Highlighting

DEPARTMENT OF COMPUTER SCIENCESOFTWARE ENGINEERING, GRAPHICS, AND VISUALIZATION RESEARCH GROUP

• Hovering over a deep node reveals the subtrees containing that node by outlining corresponding nodes

• Node and subtree highlighting combined allow the user to see multiple connections without introducing edges

Page 10: DEPARTMENT OF COMPUTER SCIENCE SOFTWARE ENGINEERING, GRAPHICS, AND VISUALIZATION RESEARCH GROUP 15th International Conference on Information Visualisation

Parent Display

DEPARTMENT OF COMPUTER SCIENCESOFTWARE ENGINEERING, GRAPHICS, AND VISUALIZATION RESEARCH GROUP

• The parents of the node are displayed in a crust around the outer edge of the largest circular node similar to the crust of a pie

• Parent nodes have same functionality as child nodes – e.g. navigation, highlighting and labeling

Page 11: DEPARTMENT OF COMPUTER SCIENCE SOFTWARE ENGINEERING, GRAPHICS, AND VISUALIZATION RESEARCH GROUP 15th International Conference on Information Visualisation

Incremental Exploration

• Designed for large systems• Graph may be too large to display in its entirety

while still maintaining a high level of detail• Viewed from perspective of a particular node• Select nodes of interest to navigate through

graph• Degree of Interest (DOI)

• Users have flexibility in setting up the level of suppression based on their DOI

DEPARTMENT OF COMPUTER SCIENCESOFTWARE ENGINEERING, GRAPHICS, AND VISUALIZATION RESEARCH GROUP

Page 12: DEPARTMENT OF COMPUTER SCIENCE SOFTWARE ENGINEERING, GRAPHICS, AND VISUALIZATION RESEARCH GROUP 15th International Conference on Information Visualisation

History

DEPARTMENT OF COMPUTER SCIENCESOFTWARE ENGINEERING, GRAPHICS, AND VISUALIZATION RESEARCH GROUP

• History nodes are added to the visualization as graph is traversed

• Provides user the ability to see where they have been, and how that relates to where they are going

Page 13: DEPARTMENT OF COMPUTER SCIENCE SOFTWARE ENGINEERING, GRAPHICS, AND VISUALIZATION RESEARCH GROUP 15th International Conference on Information Visualisation

User Study

DEPARTMENT OF COMPUTER SCIENCESOFTWARE ENGINEERING, GRAPHICS, AND VISUALIZATION RESEARCH GROUP

• 8 participants: 5 technical and 3 non-technical

• 24 questions, objective and subjective

• Objective questions: had a definitive correct answer

• E.g. browsing to a given node, counting or locating specific information

• Subjective questions: open to interpretation

• E.g. identify the most popular account in a social network

• Participants were asked to provide an explanation for their reasoning in addition to their answer

• Evaluate PieVis in situations which require critical thinking and problem solving

Page 14: DEPARTMENT OF COMPUTER SCIENCE SOFTWARE ENGINEERING, GRAPHICS, AND VISUALIZATION RESEARCH GROUP 15th International Conference on Information Visualisation

Results

DEPARTMENT OF COMPUTER SCIENCESOFTWARE ENGINEERING, GRAPHICS, AND VISUALIZATION RESEARCH GROUP

Tabulated results from the user study. Tasks have been grouped into five types; browsing, content, search, counting and subjective. Ratings marked 'x' indicate tasks that were incomplete or incorrect.

Page 15: DEPARTMENT OF COMPUTER SCIENCE SOFTWARE ENGINEERING, GRAPHICS, AND VISUALIZATION RESEARCH GROUP 15th International Conference on Information Visualisation

Results

DEPARTMENT OF COMPUTER SCIENCESOFTWARE ENGINEERING, GRAPHICS, AND VISUALIZATION RESEARCH GROUP

• Browsing and searching related tasks were rated highest• Average rating of approximately 4.5 (browsing) and 4.7

(searching).• These results are very good as they are the main areas of

interest for an incremental exploration tool such as PieVis

• Tasks that required contextual information were rated the lowest with an average rating of approximately 2.8

• This is an expected result as PieVis does not attempt to visualize content

• This area is well handled by existing methods which PieVis is meant to supplement.

• Overall the study was very successful with an average task rating of 4.15/5

Page 16: DEPARTMENT OF COMPUTER SCIENCE SOFTWARE ENGINEERING, GRAPHICS, AND VISUALIZATION RESEARCH GROUP 15th International Conference on Information Visualisation

DEPARTMENT OF COMPUTER SCIENCESOFTWARE ENGINEERING, GRAPHICS, AND VISUALIZATION RESEARCH GROUP

CONCLUSION

• We interact with large/infinite graphs everyday

• Internet

• Social networks

• File systems

• Common applications provide a view of only one node at a time

• Single webpage in a browser

• Directory in a file explorer

• One Twitter/Facebook/etc account

• PieVis provides a visualization system for exploring graphs of any size in real time

• See relationships between nodes

• Understand current position as it relates to the rest of the graph

Page 17: DEPARTMENT OF COMPUTER SCIENCE SOFTWARE ENGINEERING, GRAPHICS, AND VISUALIZATION RESEARCH GROUP 15th International Conference on Information Visualisation

DEPARTMENT OF COMPUTER SCIENCESOFTWARE ENGINEERING, GRAPHICS, AND VISUALIZATION RESEARCH GROUP

QUESTIONS

Contact and Research Information

Dr. Adrian Rusu: [email protected]://elvis.rowan.edu/~rusu

Bryan Petzinger: [email protected] Crowell: [email protected] Fabian: [email protected]

SEGV Research Group Homepage: http://cs.rowan.edu/segv

Thank you!