Partition Clustering for GIS Map Data Protection

One of the main research issues of digital data is defined by copyright protection, and digital watermarking is a potential solution to this issue. While there is an abundance of research on digital watermarking for image data, there is far less research on digital watermarking for vector map data, a data format used to store complex information in Geographical Information Systems (GIS). Recently, data mining methods have been used in the process of watermarking vector data. In this paper, we argue that the security of the watermarked vector maps can be increased by employing more suitable data mining methods. In particular, in this paper, we advocate the use of k-medoids partition clustering and compare its deployment with a previous watermarking scheme in which k-means partition clustering is used. The experimental results show that it outperforms the approach based on k-means according to a set of evaluation metrics.