Date: Thu, 21 Nov 1996 22:13:07 GMT Server: Apache/1.1.1 Content-type: text/html Content-length: 7999 Last-modified: Wed, 02 Oct 1996 01:02:07 GMT
Scott received his undergraduate degree from the University of North Carolina at Chapel Hill in 1992. His major was mathematical sciences with an emphasis in computer science. For his honors project he helped design and implement PGOMS, an interface for creating GOMS models.
Scott has worked at Oak Ridge National Laboratories (ORNL) with Tom Rowan as a member of the Visual and Information Sciences Group. While at ORNL, Scott built the first version of Irwin, an information resource monitoring and display tool, as part of the GO-NII Collaborative Tools project.
If you are interested, please examine Scott's resume.
Internet Information Monitoring and Display is an important area considering the constantly growing and changing nature of the Internet. Resources such as Email, Usenet news, and the World Wide Web provide a continuous flow of information. While users can access this information using any of a number of readers and browsers, it is difficult to determine when and if a resource needs to be accessed. In addition, the large amounts of information that accumulate can be difficult to assimilate at a glance.
Visualizations are needed that will provide an overview of these resources with a detailed look at recent changes. As a result, information about the names and content of the messages and documents must be encoded in a small number of pixels through the use of color, intensity, size, and shape. These techniques are reflected in Irwin, a tool that monitors resources and provides various alerts when changes occur. A multi-layer view of the resources simultaneously provides a broad overview of all resources plus a narrow look at recent changes without losing context.
Recently, Scott was also part of a team that developed SQWID, a Java tool for visualizing the results from an Alta Vista search. SQWID provides a stress-based graph view of the search result sites that will hopefully help users identify interesting Web sites.
Program Visualization examines methods for visualizing information about a computer program. For years programmers have drawn pictures such as flowcharts or control-flow diagrams to help other programmers and users better understand their programs. Unfortunately, these pictures will not provide the exact desired perspective for every user. Clearly, methods are needed for automatically generating graphical representations of programs.
Existing code formatters, profilers, and analysis systems provide adequate static and post-mortem views of a program. However, debugging and understanding is aided if the program execution is shown in conjunction with the program code. By visually connecting the code and execution, the user can identify elements in the program that otherwise would not be visible. Scott is working on RunView, a runtime-based code analysis system that incorporates these and other ideas. He is also exploring how RunView can be used to decompose software architectures in a project with Gregory Abowd.
Software Visualization is the use of graphics, visualization, and animation as an aid in the understanding of algorithms, architectures, and software. Software visualizations can be constructed using toolkits like Tango and Polka. With these toolkits, a programmer creates objects and programs them with actions that occur over a number of frames.
With the advent of improved hardware, the number of frames displayed per second is sure to decrease. Thus, a natural evolution is to specify actions not in frames but in real time (seconds and milliseconds). This concept is central to Polka-RC (Polka Real-Clock), an evolution of the original Polka system with the added capability of time-based animation activations and durations. In addition, Polka-RC provides more natural ways to describe actions and their relationship to objects. Tech report GIT-GVU-95-21 discusses these ideas in detail.
Wait-Free Consensus reflects the ability of N distributed objects to ``agree'' in the presence of N-1 failures. A number of object parameters can impact on this ability to agree. See GIT-CC-94-04 for a discussion of these parameters and a look at the properties of some distributed objects.