RESEARCH GROUPS DETAIL

Information Security, Data Mining & Business Intelligence Research Group

Introduction

We are living in a digital world where tera-bytes of data is generated daily all around the world. This data cannot be useful until and unless some useful knowledge is extracted from it. ISDMBI is focused on doing quality research for developing novel ideas for extracting useful hidden knowledge in the data. This knowledge can further be used to propose different solutions particularly when dealing with large datasets. Moreover, when experts from different domains need to work in collaborations, the data needs to be shared among different stakeholders. This sharing may cause serious concerns related to data security e.g. data privacy, data ownership etc. In this context, ISDMBI is also focused to address such security concerns. In a nutshell, research at ISDMBI is specialized in the areas: data mining, machine learning, data warehousing and decision support system, fingerprinting, differential privacy and watermarking.

Group Name: Information Security, Data Mining & Business intelligence Research Group
Team Lead: Dr. Muhammad Kamran
Group Secretary: Arubah Hussain

Research Area:

The ownership of digital data is one aspect of data security. Design and development of knowledge-preserving and robust watermarking techniques for digital ownership rights protection of relational databases is an active area of research. In this research, various disciplines have been used to design and develop watermarking techniques which ensure knowledge-preserving and robustness characteristics of watermarking 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 mining 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 identify 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, we working on techniques to make the watermark decoding accuracy independent of this bandwidth and hence the usability constraints; as a result, maximum 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 preserved 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.

Group Members:

Dr. Kamran
Dr. Muhammad Javed Iqbal
Mr. Majid Mumtaz
Mr. Amjad Usman
Ms Arubah Husain