Domain-Specific Model Distance Measures

A lot of research was invested in the last decade to develop differencing methods to identify the changes performed between two modelversions. Typically, these changes are captured in an explicit difference model. However, less attention was paid to quantifying the distance between model versions. While different versions of a model mayhave the same amount of differences, their distance to the base model may be drastically different. Therefore, we present distance metrics formodels. We provide a method to automatically generate tool support for computing domain-specific distance measures. We show the benefits of distance measures over model differences in the use case of searching for the explanation of model evolution in terms of domain-specific change operations. The results of our experiments show that using distance metrics outperforms common difference models techniques.