Automated use case similarity computation can aid the assessment cohesion and method complexity of classes

Use cases are widely used in early software development phases such requirements analysis. In this paper, we investigate how use case similarity could impact the classes that implement them. We studied if the similarity of use cases could have impact on the lines of code shared between them and on metrics of classes that implement them, such as, coupling, cohesion and method complexity of classes. We also have successfully applied an automated approach to assess the similarity of use case names. We found that there is a statistically significant correlation, altough not strong, between use case similarity and sharing of lines of code. Interestingly, we also have found that classes that are shared between different use cases tend to have lower cohesion. Moreover, classes that are shared between similar use cases tend to have higher method complexity in classes. We found no relation between use case similarity and coupling. Keywords—Use Case Similarity, Software Metrics, Software Maintenance