Wertebasiertes Requirements Engineering am Beispiel medizinischer Software
Zusammenfassung der Projektergebnisse
The project VaReMed focused on the relation between personal values of software users and the software features they like. The overall goal is to support user satisfaction with software at the workplace by reflecting personal values during requirements engineering. This project has performed five quantitative and qualitative studies to investigate the relation between values, user tasks and software features as well as to investigate measures for predicted user satisfaction. Three studies were performed in the context of medical software (used by physicians or nurses), as various professions are involved as users and the quality of the software is critical. In parallel, fragments of a method for using values during requirements engineering were developed. The results of the project can be summarized as follows: The two questionnaires known from literature to elicit personal values worked well in the medical domain. However, they are too intrusive to be used at the workplace during requirements engineering. It is less intrusive to elicit preferences for work tasks. Our studies showed that a questionnaire on preferences of work task can serve as proxy for the personal value questionnaires in a specific domain. Small differences between holders of different values as to preferences for already existing features of their clinical software systems encouraged to pursue value dependence of features. However, statistical evidence for relations between personal values and preferences for software features could not be shown in the medical studies. One reason might be that this project studied quantitatively only the relation between personal values and predicted user satisfaction. Further quantitative domain-independent studies confirmed that care has to be taken when measuring predicted satisfaction, as e.g. this is also influenced by familiarity with and actual understanding of the described features. The studies in the medical domain confirmed values of physicians and nurses already known from literature. Furthermore they showed that personal values matter in the workplace and that a large percentage of clinical professionals concentrate at few dominant values. Therefore, it is meaningful to strive for medical professions reference models and to target software at the demands of the respective values. A method for using values during requirements engineering can rely on a reference model of relations between preferences for work tasks and preferences for software features in a given domain. However, developing such a reference model for each domain requires a lot of effort. A domain-independent reference model between general tasks and features which can be instantiated for different domains has been developed, but not evaluated.
Projektbezogene Publikationen (Auswahl)
-
(2010) Use of Personal Values in Requirements Engineering- A Research Preview, Conference on Requirements Engineering: Foundation for Software Quality (RefsQ'10), LNCS 6182, pp. 17-22, Springer
Proynova R, Paech B, Wicht A, Wetter T
-
(2010): What if “business process” is the wrong metaphor? Exploring the potential of Value Based Requirements Engineering for clinical software. Safran C, Reti C, Marin HF (eds) MEDINFO 2010 - Proceedings of the 13th World Congress on Medical Informatics. Amsterdam (IOS), p 1389
Wetter T, Paech B
-
(2011) Approximating user values to preserve privacy - a proposal. 1st International Workshop on Values in design (VinD'11 ), In conjunction with the 13thIFIP TC13 Conference on Human-Computer Interaction
Koch SH, Proynova R, Paech B, Wetter T
-
(2011): Can we improve clinical software based on users’ personal values? Identification of potentially beneficial software features. Proc. AMIA Annual Symposium 2011 p 1838
Koch SH, Proynova R, Paech B, Wetter T
-
(2012): Do Stakeholders Understand Feature Descriptions? A Live Experiment. Post-Proceedings REFSQ 2012, pp. 265-280, ICB_Report 52, University of Duisburg-Essen
Proynova R, Paech B
