Directed Research Project Details
Simulating Specific Users For Software Testing
| Author: | Darren Brown |
|---|---|
| Date: | December 13, 2004 |
| Time: | 10:00 |
| Location: | 220 Deschutes |
| Committee: | Steve Fickas (Chair) Art Farley Michal Young |
Abstract
The software testing problem exists simply because it is impossible to completely test software with finite resources. Statistical testing techniques such as Operational Profiling attempt to mitigate this by using conglomerate user data to create a simulation model. We contend that such an approach may not accurately portray software use, since it undermines the heterogeneity of software environments. We suggest augmenting such statistical techniques by identifying and simulating specific classes of users in hopes of finding software defects that would otherwise be found by users. We begin our consideration by modeling a particular user who has found numerous software bugs in the past. We simulate this model to find specific, real-world bugs and compare this approach to random simulation.