On deriving conceptual models from user requirements: An empirical study

Context: There are numerous textual notations and techniques that can be used in requirements engineering. Currently, practitioners make a choice without having scientific evidence regarding their suitability for given tasks. This uninformed choice may affect task performance. Objective: In this research, we investigate the adequacy of two well-known notations: use cases and user stories, as a starting point for the manual derivation of a structural conceptual model that represents the domain of the system. We also examine other factors that may affect the performance of this task. Methods: This work relies on two experiments. The first is a controlled classroom experiment. The second one is a quasi-experiment, conducted over multiple weeks, that aims at evaluating the quality of the derived conceptual model in light of the notation used, the adopted derivation process, and the complexity of the system to be. We measure quality in terms of validity and completeness of the conceptual model. Results: The results of the controlled experiment indicate that, for deriving conceptual models, user stories fit better than use cases. Yet, the second experiment indicates that the quality of the derived conceptual models is affected mainly by the derivation process and by the complexity of the case rather than the notation used. Contribution: We present evidence that the task of deriving a conceptual model is affected significantly by additional factors other than requirements notations. Furthermore, we propose implications and hypotheses that pave the way for further studies that compare alternative notations for the same task as well as for other tasks. Practitioners may use our findings to analyze the factors that affect the quality of the conceptual model when choosing a requirements notation and an elicitation technique that best fit their needs.