Building Markov chain-based software reliability usage model with UML

Several software usage models based on Markov chain have been presented in literature in the last three decades. However, existing software usage models have the common problem that the researchers directly give the value of transition probability, but not give a method for determining the transition probability. Use case and scenario-level usage model have been created based on extended UML model. In the process of scenario-level usage model, this paper mainly focuses on the determination of transition probability and analysis of the relationship between use case and scenario-level usage model. We introduced the improved AHP method to determine the transition probability. AHP has the advantage that even if not familiar with the software, we can still determine the transition probability more accurately. Finally, we separately devise a algorithm to achieve the process of creating Markov chain use case and scenario-level usage model from UML model.