Abstract:
Design and development of knowledge-preserving and robust water-
marking techniques for digital ownership rights protection of relational
databases is an active area of research. In this dissertation, various dis-
ciplines have been used to design and develop watermarking techniques
which ensure knowledge-preserving and robustness characteristics of wa-
termarking techniques by bringing tolerable distortions in the original
data. Different parameters are used to define data distortions in terms
of information loss as a result of watermarking. The intelligent min-
ing techniques and statistical measures have been utilized to define and
measure information loss after watermark embedding in the relational
databases. The data owner usually defines the usability constraints to
control this information loss. These usability constraints in turn iden-
tify the available bandwidth for watermark embedding. The watermark
decoding accuracy of a watermarking algorithm generally depends on
this bandwidth; larger the bandwidth the better the decoding accuracy
(watermark robustness) and vice versa. However, this dissertation pro-
poses a model to make the watermark decoding accuracy independent
of this bandwidth and hence the usability constraints; as a result, max-
imum decoding accuracy can be achieved even with very tight usability
constraints and minimum data distortions. Such mechanism also helps
to preserve the knowledge in the databases to a maximum level; as a
consequence, the classification results for such databases are also pre-
served after watermark embedding. In the pilot study, empirical study
and formal modeling have been used to prove the knowledge-preserving
and robustness characteristics of the proposed watermarking techniques
for digital right protection.
