Being confronted with a defect in software leads to the well known task: correcting the software such that the defect does not occur anymore. Here, the location of the defect and the corresponding root cause do not have to be identical. Thus, before any correction can be done, the reviewer has to detect the root cause for the defect. In order to reduce the reviewing effort, this paper presents a method to automatically narrow down possible root causes for defects found in Simulink models. Starting at a defect location, a backward search is applied to detect all paths leading to that defect. Each path is weighted by previously determined weights depending on the block types contained in the respective path. This weighting correlates with the probability of a path containing the root cause.

Being confronted with a defect in software leads to the well known task: correcting the software such that the defect does not occur anymore. Here, the location of the defect and the corresponding root cause do not have to be identical. Thus, before any correction can be done, the reviewer has to detect the root cause for the defect. In order to reduce the reviewing effort, this approach presents a method to automatically narrow down possible root causes for defects found in Simulink models. Starting at a defect location, a backward search is applied to detect all paths leading to that defect. Each path is weighted by previously determined weights depending on the block types contained in the respective path. This weighting correlates with the probability of a path containing the root cause.