Strengthening data sovereignty through digital text watermarking

Abstract

With the continuous shift from the analog to the digital world, data has become a common resource in our everyday lives. Individuals care about protecting their personal data, and companies keep data secure to gain a competitive advantage. Nevertheless, sharing and using data in an interconnected world is necessary to create value through cooperation. Various international regulations increase the complexity of the tension between data sharing and data protection. Keeping control over data, often referred to as data sovereignty, is of utmost importance. However, the concept of data sovereignty is interpreted differently and lacks a precise delimitation and clear conceptualization. Existing technical solutions for maintaining data sovereignty remain limited and specialized.

This cumulative doctoral dissertation uses Design Science Research to address the lack of technical solutions for strengthening data sovereignty. First, we conceptualized and delimited data sovereignty from adjacent terms in the domains of information systems and software engineering. We reviewed the existing literature and conducted interviews with practitioners, thereby identifying data sovereignty challenges and highlighting digital watermarking technologies as a promising solution approach. Second, we designed and developed a digital text watermarking artifact, Innamark, to mitigate the identified data sovereignty challenges. It focuses on protecting text, the most widely shared and used content type, with limited watermarking solutions available. Third, we derived design knowledge in the form of design principles to generalize our findings based on our experience and gathered insights.

We instantiated Innamark as an IT artifact, including a reusable Kotlin multiplatform library. Different demonstrations, in the form of implemented applications and a test in a data space context, supported us during the iterative design process. Further, we evaluated Innamark with respect to embedding capacity, invisibility, and robustness. Using a benchmark evaluation on a dataset of 1 000 000 articles, we compared Innamark with nine related algorithms. To the best of our knowledge, the findings show that Innamark is the first invisible and robust watermarking method that does not increase the length of the watermarked text, despite its limited embedding capacity.

Our three contributions (i) the conceptual model for data sovereignty, (ii) Innamark as a digital watermarking artifact, and (iii) the resulting design principles, support a broad range of applications and help researchers and practitioners to strengthen data sovereignty.