Join us for a PhD thesis defence of Bjarki Freyr Sveinbjarnarson on his thesis: Scaling Data Governance Through Automation and Dynamic Data Spaces.

Defence committee:

Main Supervisor: Anna Sigríður Islind, Associate Professor, Department of Computer Science, Reykjavik University, Iceland.

Co-Supervisor(s): Erna Sif Arnardóttir, Associate Professor, Department of Computer Science & Department of Engineering, Reykjavik University, Iceland.

Committee:

• Tomas Lindroth, Assistant Professor, Department of Applied IT, Gothenburg University, Sweden.
• Stefán Ólafsson, Assistant Professor, Department of Computer Science, Reykjavik University, Iceland.

Examiner:

• Olivia Benfeldt, Assistant Professor, Department of Digitalization, Copenhagen Business School, Denmark.

Master of Ceremony: Luca Aceto

Abstract:

Managing vast amounts of heterogeneous data is a growing challenge in healthcare research, where data must be structured, stored, and validated for diverse research purposes while minimizing the manual resources required for maintenance. This thesis explores scalable data management solutions, focusing on dynamic data spaces, data quality enforcement, and automated data governance to ensure usability across research disciplines. 

We designed and developed a homogeneous database structure capable of accommodating a wide range of structured health data while minimizing complexity and ensuring flexibility for future research needs. This architecture enables the seamless integration of new data sources without expanding storage requirements, making it a scalable solution for large-scale research projects. Additionally, we implemented a validation system embedded within a digital infrastructure to enforce predefined data standards, reducing errors at the point of entry and improving data quality. 

The empirical work underlying this thesis is based on four studies conducted within a large sleep research project encompassing diverse datasets. First, we evaluated the feasibility of displaying key dataset insights through a digital infrastructure, providing researchers with an overview of stored data. Second, we developed a homogeneous database framework that eliminates the need for additional tables as new data sources are added. Third, we conducted a study to identify the root causes of poor data quality at the input stage, analyzing how researchers interact with data validation mechanisms and where errors emerge. Lastly, we examined data governance models that minimize human intervention while maintaining research integrity, ensuring sustainability in data management. 

Findings from these studies reveal common causes of poor data quality, including a lack of metadata prioritization, reliance on assumptions instead of documentation, inconsistent formatting, and diverse interpretations of what constitutes as good data. The findings also illustrate conflicting priorities regarding data governance, further complicating standardization efforts. Despite these challenges, our homogeneous database design and automated data governance framework provide a workable solution for multidisciplinary research projects by reducing manual oversight and improving long-term data usability. This thesis contributes to the fields of data governance, database architecture, dynamic data spaces, information systems, and digital infrastructure by demonstrating how scalable, automated solutions can streamline data processes while ensuring high-quality, standardized data for diverse purposes.