Abstract
Computer systems that observe user behavior in order to detect malfunctions and respond to them are mostly unexplored from a scientifi c point of view until now. However, they may hold much potential to improve human-computer interaction. This thesis proposes a reference model for such systems. It describes common computer problems and their symptoms, reactions by the users and ways to respond to such problems on system side. Taxonomies covering common use-cases are provided and both psychological and technical aspects are covered.
In order to show the effectiveness of behavior-based malfunction response, three systems were implemented: the Microphone Helper assists the user in confi guring the microphone. The Mouse Helper recognizes confused user behavior in situations of mouse cursor freezes.
The Frustration 2.0 project tries to prevent user frustration when encountering buggy JavaScript-based websites and provides the manufacturer of the website with diagnostic information. The first two projects were evaluated empirically in user studies: the detection rate and the time gains were measured and the opinions of the users about such systems were analyzed. The results indicate that behavior-based malfunction response can be helpful in many cases, but they also show speci fic problems that have to be considered when designing such systems.
The full diploma thesis (PDF, ~2MB)