University of Twente Student Theses


Identifying unwanted requirements dependencies: developing a method for requirements dependency modeling

Vliegen, Laurens (2012) Identifying unwanted requirements dependencies: developing a method for requirements dependency modeling.

Full text not available from this repository.

Full Text Status:Access to this publication is restricted
Abstract:In this thesis, a method for identification, classification and modeling of dependencies between functional software requirements, and identification of unwanted dependencies is proposed. Unwanted dependencies are dependencies which might cause problems during later stages of the project, especially in the form of work dependencies. Managing work dependencies is important because development activities involved with interdependent requirements should be coordinated in order to maintain efficiency in the project. Failing to do so could decrease the efficiency of project activities, and could result in an end-product that does not meet the stakeholders' requirements. The basis of the developed method is formed by a classification for functional software requirements dependencies, which was derived from literature. The classification was incorporated in an initial requirements modeling method, and applied to two requirements specifications from real life projects. This validation provided feedback on both the classification and the modeling method, which was then processed into a revised method. Because of the resource-intensiveness of application of the full method, the revised method included guidelines for assessing the priority of the dependency types in the classification, providing a light weight approach to the method. The required resources can thereby be limited by identifying only the most important or dangerous dependencies in a requirements specification. The classification was then mapped to the classification as part the Archimate standard, for modeling motivations behind enterprise architectures. This comparison and the guidelines for determining the priorities of the dependency types were finally validated through an interview with BiZZdesign, whose Architect tool uses the Archimate standard for modeling enterprise architectures and their requirements. The first validation process provided insights into the ambiguities of the method and classification, and especially regarding the subjectivity involved in the application of the method and classification. The lack of objectiveness differs per dependency type, and thereby causes the resource-intensiveness of the identification of the dependency types to vary drastically. From the first and second validation, the author's professional experience and experts from the field of requirements engineering, it became clear that the quality of requirements specifications is overestimated, and that both this poor quality and the dependencies in the respective specifications can lead to serious problems during later stages of the project. The second validation process furthermore confirmed the results from the first validation regarding the poor quality of the respective specifications, and the subjectivity involved in application of the method. It furthermore provided more insight into the differences and commonalities between the compared classifications. These can mainly be attributed to the differences in perspective and use context. The main conclusions to this research are that the developed method provides a solution to the problems surrounding unidentified requirements dependencies, and the classification provides a solid basis for use in requirements modeling tools. The method should be used by requirements engineers, for improving the quality of requirements specifications before these are passed on to the developers. We can recommend BiZZdesign to use the proposed classification when developing a tool for modeling functional software requirements, which is a business case currently under investigation. Furthermore we can conclude that, based on the second validation process, that the proposed classification is applicable to both high-level modeling situations, as in the Architect tool, as well as low-level modeling situations, as in the proposed method. This is an interesting finding, because it is generally assumed that classifications for objects on different levels of abstraction will contain different types of relations
Item Type:Essay (Master)
Faculty:BMS: Behavioural, Management and Social Sciences
Subject:85 business administration, organizational science
Programme:Business Information Technology MSc (60025)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page