Extended Trace-based Task Tree Generation

Task trees are a well-known way for the manual modeling of user interactions. They provide an ideal basis for software analysis including usability evaluations if they are generated based on usage traces. In this paper, we present an approach for the automated generation of task trees based on traces of user interactions. For this, we utilize usage monitors to record all events caused by users. These events are written into log files from which we generate task trees. The generation mechanism covers the detection of iterations, of common usage sequences, and the merging of similar variants of semantically equal sequence. We validate our method in two case studies and show that it is able to generate task trees representing actual user behavior.