Accessing relational databases through web services increases risks of piracy and tampering. Watermarking guarantees data protection without constraining their access and management. Researchers traditionally compare the extracted watermark with the embedded one after simulating attacks to analyze its robustness. However, the literature has not addressed how much watermark signal is retained after querying multi-word textual attributes and how topological features of the watermark affect robustness. In this work, we fill this gap by evaluating a semantics-preserving approach. We execute different classes of queries in three data sets, including an actual database of a public administration. The results show that 99.99% of the marks can be detected when considering a watermark source with a topological factor close to or equal to 1. As valuable contributions, we formally defined the watermark source’s topological factor and the corrosion degree of the watermark detection. This work offers hints to data owners to adjust detection parameters, increasing the chances of spotting the watermark despite having, as evidence, data highly damaged due to malicious operations.
Distortion-based watermarking techniques embed the watermark by performing tolerable changes in the digital assets being protected. For relational data, mark insertion can be performed over the different data types of the database relations’ attributes. An important goal for distortion-based approaches is to minimize as much as possible the changes that the watermark embedding provokes into data, preserving their usability, watermark robustness, and capacity. This paper proposes a quantile-based watermarking technique for numerical cover type focused on preserving the distribution of attributes used as mark carriers. The experiments performed to validate our proposal show a significant distortion reduction compared to traditional approaches while maintaining watermark capacity levels. Also, positive achievements regarding robustness are visible, evidencing our technique’s resilience against subset attacks.