The impact of use case format on the understanding of system requirements by novice and experienced users

Abstract

The effective analysis and specification of requirements is critical in software development. Faults in the requirements may later have significant impact on the quality of the software system. Ineffective communication between users and developers is a major cause of failures of software projects. Use case model is a powerful and widely recognized tool for elicitation and specification of functional software requirements in object oriented methodology. It has been advocated as a way to negotiate and communicate requirements between system analysts and stakeholders. However, issues concerning the format, level of details and the communication capability of use cases are still unclear and debatable. This study uses theories from cognitive psychology on human understanding to derive hypotheses on the effect of the format of use case model on user understanding. In this study, comprehension of the functional requirements are compared between experienced and novice users. Particularly, the effect of differences in use case format on novice and experienced users performance in both familiar and unfamiliar domains were explored and if combining the textual description of a use case with diagrams of different levels of detail improves their understanding. Two controlled experiments were conducted; one to assess the performance of novice users, the other to assess more experienced users. The results of both experiments provide evidence that support the propositions that individuals who view text with use case diagram (simple or detailed) will develop higher level of understanding of the system requirements in less time when compared to individuals who view a text only model. The results of both experiments provide no evidence that support the propositions of the benefit of the simple diagram for improving novice users understanding, and the detailed diagram for aiding experienced users when combined with the text description. It is also found that neither the observed level of prior domain knowledge nor the observed level of analysis method knowledge has a significant effect on the level of “understanding” that users developed regarding a system requirements. Finally, our analysis shows no considerable differences in performance in the experiments tasks between novice and experienced users, which mean that the effect of experience on users understanding is still an open issue and needs further research in the future