Utility classes detection metrics for execution trace analysis

Execution trace analysis is particularly valuable in the context of object-oriented software comprehension for the maintenance tasks. It involves analyzing dynamic information and behaviors that are stored in the execution trace. However, the execution traces of current object-oriented software systems tend to be very large in terms of complexity and size. In particular, classes and coupling features form a very complicated interwoven lattice of the dynamic dependencies. This research focuses on detecting and removing utility classes from an execution trace. Utility detection and removal techniques are very useful in reducing the structural complexity of execution traces for software comprehension process. For this purpose, two new utility classes detection metrics based on dynamic coupling have been introduced to measure the extent to which a class can be considered as a utility.