Discussion questions from CS 7390 lecture

Date: 1/12/96

Presented by: John Stasko

Topic: Two taxonomies of software visualization

• "Intention Content" is a curious term. Can you explain it and its importance in a bit more detail?
• Is there any way to have automation and high intention content at the same time?
• Automation seems to be independent of what a software visualization system does or is. ie. automation can be added a priori if a system is successfull. Why make a distinction in this term?
• How does the description of automation by Stasko & Patterson differ from the description of specification method by Roman & Cox?
• Is there a more objective formal definition of animation? What do the terms "sufficient" and "transform continuously" mean in this context?
• How is animation a continuous dimension if, by definition, a visualization is either animated or not animated.
• Couldn't "rapid display" of valid states imply animation without any "in between" states, e.g. an accelerated animation of a thermometer?
• Do you know of any dangers in showing in-between states that might confuse rather than help the user?
• In Stasko's paper, for the animation part, how is the in-between generated using linear interpolation or some other techniques?
• Can annotation be done in such a way that it can easily withstand changes in program code/structure?
• How does Stasko's & Peterson's taxonomy account for or deal with the user and, in particular, user interaction with the visualization system? This seems to be the main difference in the taxonimies in the 2 papers. If it doesn't deal with user interaction, is there a reason why?
• Roman & Cox's paper and taxonomy is based on the mapping of a program to a graphical representation. It discounts all other taxonomies because they are not based on well-formulated models in software visualization. But, the authors give no basis or support for their "model". In fact, they barely discuss their mapping model. Is their mapping model a widely accepted model in software visualization? Is this why there was no defense for their model campared to other models?
• In Roman's paper, the author claims that the taxonomy is derived directly from the formal model of program visualization as a mapping. How is the taxonomy connected with the model?
• Roman & Cox lay out 5 criteria for evaluating software visualization systems. To what extent are the other four limited or pre-defined by the choice of scope?
• What benefits does software visualization provide for the end-users of large software systems? In particular, how can software visualization be used to assist end-users of database systems, on-line transaction processing systems, or command, control, and communication (C3) systems where the end- user must often wait while the software system performs some task?
• How might software visualization appropriateness be determined? What information would be required about the goals/needs of the intended user (programmer, student, end-user, etc.) and the capabilities of the software to be visualized?
• What is meant by "analytical presentation"?
• What is declaration (in the specification method of Roman & Cox, pg. 19)
• When you start making these techniques interactive, doesn't (or shouldn't) the boundary between visualization & visual programming become much less definite? Should the programmer have to go back to his code to make changes that come up in the visualization?
• What in software visualization is different from data or information visualization? Is it a subclass, a mixing of both, or ....? And what in the taxonomy is specific to software visualization (vs. data or info vis)?
• How do you relate/differentiate aspect, abstraction, and automation (of Stasko and Patterson) with Scope, abstraction, and specification method (of Roman and Cox)?
• What are the similarities/differences between the various taxonomy classification dimensions mentioned in the papers? Why are some chosen over others? Why is one better than another?
• Have any studies been performed to gauge the effectiveness of algorithm animation as learning aids, and if so, what are the results?
• Are the four dimensions suggested by Stasko and Patterson adequate to classify the exisiting software visualization systems, or are the five dimensions suggested by Roman and Cox necessary?
• For non-imperative languages (functional, logic, etc.) what are the important programming aspects and how are those visualized?