A pattern based approach for data quality requirements modelling

Abstract

Data quality (DQ) is a critical issue in today’s information systems. Both academic researchers and industry practitioners have contributed to addressing the problem of data quality through management strategies and technological advancements over the last three decades, yet data quality management remains a challenge in the organizational management portfolio. Requirement models have been used successfully to develop solutions in areas like software and database development. The current state of the art in DQ management methodologies developed by both academic researchers and industry practitioners have largely overlooked the area of DQ requirements modelling and analysis. However DQ requirements are fundamental to DQ management since the ultimate purpose of DQ management is to develop solutions to satisfy the DQ requirements. Thus, a clearly defined DQ requirement model is a necessary prelude to systematically develop solutions to organizational DQ problems. In this research, we have developed a repository of thirty-three DQ patterns to model DQ requirements. The patterns are rich in representing the real world DQ requirements while free from notational complexities, thereby allowing them to be used practically to support DQ management. We used design science as the guiding methodology for developing DQ patterns while maintaining a rationale for the rigor and the relevance of our artefacts through appropriate validations and verifications throughout the design process. One of the challenges faced in the conceptualization of DQ patterns was the lack of shared understanding among researchers about DQ dimensions, which is a key concept in representing a DQ requirement. Owing to the importance of shared understanding we systematically refactor the concept of DQ dimensions by consolidating different viewpoints from both academic and practitioner community. As a secondary aim of this study, we adapted a credible requirements engineering methodology from literature to analyse and elicit DQ requirements. We demonstrate through empirical studies that by using this methodology DQ patterns can be effectively used to elicit and model DQ requirements in organizations.