C SCI 4600 - In-Class Exercise No. 3

C SCI 4600 -- The Human-Computer Interface
Spring Semester, 2002

IN-CLASS EXERCISE No. 3
Tuesday, February 5, 2002

The Goal: To develop visualizations for large data sets which help users explore for items of interest.

The Scenario: An important way in which computers can enhance our productivity in certain tasks is by presenting numeric data, as well as other kinds of information, in alternative forms which are easier for us, as people, to deal with. In scientific visualization the objective is to help users detect patterns in the data; in information visualization the goal is to help users understand the structure of the data. Because in real world applications the quantity of data can be so large that it becomes impossible to display the entire data structure on the screen all at once without losing detail, a successful visualization will embody techniques that (a) help users acquire an overall (global) picture of the data they are exploring, and at the same time (b) help them retain a (local) sense of context as they navigate through the data, so they always know where they are and how they got there. In this exercise, we will focus on two important special cases of the information visualization task:

Hierarchical information, in which there are many different levels and sublevels. An example of this is a computer's tree-like file system, where to find a desired file or subdirectory we often need to move up or down along a branch, or from one branch to another. Microsoft provides the graphical Windows Explorer tool to help do this in their operating systems.
Linear information, in which we may need to search through one or more long (sorted) lists of items of multiple types in order to locate a specific item of interest. An example is a calendar of meetings and deadlines, both past and future, where we may need to recall what we did on a certain date last year, when we last met with an important client, or when a certain report will be due.

Some researchers have advocated using either very large (wall size) displays, or multiple smaller screens, to improve visualization techniques. Others have explored the use of fancy head gear and associated paraphernalia to support what is sometimes referred to as data immersion--a form of virtual or artificial reality, in which the user is led to believe that s/he is walking around inside (a graphical representation of) the data. Just how useful such approaches may be remains unresolved, and there are numerous open research issues; in any case, in this exercise we will restrict ourselves to more readily available "traditional" equipment.

What To Do: Can you come up with a pair of good techniques, one for visualizing hierarchical data and the other for visualizing linear data as described above, on today's conventional high end PCs? Divide up into small teams of 4-6 members each, and rearrange seating as necessary so as to facilitate interaction among team members. Feel free to discuss whatever you wish with me during the period, if your team encounters issues or problems you're unsure how to resolve. Towards the end of the session, I will call for volunteers from the various teams so we can compare the different designs we've come up with as a group. Finally, we will view a short video of some experimental designs implemented about a decade ago at Xerox Corporation's Palo Alto Research Center (PARC), to see how these professionals and seasoned researchers approached the same problem.

As usual, after the conclusion of the in-class part of the exercise your team should summarize its designs and the lessons learned in a written report which includes diagrams and illustrations as needed. Discuss what you perceive to be the advantages of your designs, and also possible problems. The report (one copy with all team member names listed) is due ten days after the in-class phase of the exercise is completed.