Visualization for computer security
(dev) Design new visualizations that allow network administrators to monitor large networks of computers for potential security attacks. Administrators should be able to monitor and examine communications, enabling them to spot suspicious communication patterns among the normal traffic. This could be linked to visualizations of processes/ports/users occuring on individual machines to support more detailed analysis of communications. Users of these visualizations might be administrators such as those at the VT Network Ops Center, or individual users protecting their own machines.
Advisor: Glenn Fink
Combining visualization and data mining for GIS
(dev) Combining mining and visualization can result in powerful discovery capability. Design visual and interactive methods for combining data mining results into interactive visualizations for GIS data. Our Census DataMaps visualization software could be used as a platform to explore these techniques with CT Lu's algorithms for mining spatial outliers and other patterns. How can mining results be displayed in the visualization? How can users direct the mining from within the vis?
Advisors: Qing Li, Sujatha K.
Combining visualization and data mining in Fusion (possible bioinformatics focus)
(dev) "Redescription" is a method selecting data in one context and then re-describing it into another context to understand relationships between the 2 contexts. Linked visualizations in Snap can be extended to support this type of visual analysis, and combined with Proteus (Dr. Ramakrishnan) which supports automated mining of such relationships. The "Fusion" project seeks to combine these very analogous methods. Users could interactively redescribe data in the visualizations or explore mined redescriptions. User interaction guides the mining, which in turn guides users' interaction. Design visualization, user interface and architectural methods to combine these techniques. Bioinformatics is a likely domain to explore these methods within, based on the work of the Expresso group.
Advisors: Kiran Indukuri, Qiang Yu, Sujatha K. (also Deept and Dr. Ramakrishnan)
Visualization of pathway networks in Bioinformatics
(dev) Biologists model the biological functions of organisms using network diagrams. The diagrams can be large, complex, inter-connected, and multi-scale, representing bio functions from anatomical level down to molecular level. Biologists attempt to construct, modify, and verify these diagrams based on experimental data (e.g. gene expression data), simulation, and research literature. Design new visualization display and interaction strategies for such networks that enables biologists to get an overview, drill-down, overlay experimental data, and compare alternative connections. Such a system could have profound impacts in the bioinformatics community.
Advisors: Purvi Saraiya, Dr. Karen Duca (VBI)
Information-rich virtual environments (IRVE)
(dev or eval) This project could be either development or evaluation, and multiple projects could be supported. IRVEs are useful for visualization of data that contains both 3D geometric information (VE) as well as related abstract information (InfoVis). Design and explore new strategies for visualizing such data. Alternate approaches include: methods for interactively linking VEs to InfoVis's using a Snap-like approach with X3D, or methods for visually embedding abstract data within the VE itself. Another project could explore multi-scale navigation in such spaces. Evaluation projects could evaluate such interactive strategies to determine effectiveness, or compare methods under different user tasks.
Advisors: Nicholas Polys, Dr. Bowman
more info:
Multiple-view visualization vs. integrated visualization
(eval) Visualizations can consist of a single integrated view, or of multiple coordinated views. Integrated views have the advantage of placing all of the information in a single context, the disadvantage is that the view could become too complex. Multiple views can simplify a complex visualization by dividing the information into more manageable pieces. Designing multiple view visualizations is a challenging and emerging research area. Currently, high level design guidelines exist for both integrated and multiple view visualizations. But, while integrated views have extensive low level guidelines, multiple views lack low level guidelines. Potential projects include:
Advisor: Beth Yost
Aggregation strategies for visualizing very large data tables
(dev or theory) Methods are needed to support visualizing very large data tables with 100, 1000, or more columns. A variety of projects can be done in this area:
Advisors: Qing Li, Sujatha K.
Tiled Window Manager for multiple view visualization
(dev) Design gesture-based window manager for organizing tiled windows for use in multiple view visualization systems.
Advisors: Chandresh Chhatpar, Youngyun Chungbaek
Usability of information visualization
(eval or dev) Explore usability issues of information visualization:
Advisors: Qing Li (DataMaps); Purvi Saraiya and Kiran Indukuri(bioinformatics)
Visualization of usability engineering data
(dev)
Advisors: Pardha Pyla, Chris Catanzaro
Dynamic Linking visualization software for GIS and statistical graphics
(dev) Design open-ended methods for linking commercial tools such as ESRI ArcView, SAS JMP, Spotfire, etc.
Human-eye View: controlling level of detail with eye tracking
Notification systems: controlling level of detail
...