A Cluster-Based Watermarking Technique for Relational Database

Database watermarking is an efficient method for database security. This paper proposes a cluster-based database watermarking technique which first applies cluster theory to the database watermarking technology. The cluster theory is used to cluster the source data and the clustering results determine the quantity of embedded watermark information and embedded position. This method makes the watermark information more disperse and hidden. This paper also introduces an odd-even modifying method to denote the watermarking information to decline the modification of original database. The results of experiment show this technique has a good performance on invisibility and is able to defend all kinds of subset attacks.