Protecting personal data

With expertise in data security and privacy in data mining, Professor Ljiljana Brankovic has had a long-term interest in the growing problem of protecting an individual's personal data, the threat to individual privacy and how that private information is being 'mined' and used for purposes that the individual may not approve of.

Professor Ljiljana Brankovic with the VCA Vice-Chancellor's Award for Teaching Excellence and Contribution to Student Learning in 2014 capped off a productive year for Brankovic from the School of Electrical Engineering and Computer Science. The award recognises the role of an academic from each faculty for their outstanding and diverse contribution to the quality of student learning, and Brankovic was selected as a Highly Commended Recipient among the faculty winners.

Her research in computation theory and mathematics contributes to a number of different fields but the realm of data security and privacy is one that is increasingly affecting Western society and engenders more and more attention.

With a Bachelor of Engineering (Electrical) from her native Yugoslavia and a PhD from the University of Newcastle, Professor Brankovic has cemented her position in the Faculty with significant research grants and numerous conference invitations, book chapters and journal articles. In addition to her primary research interests in graph theory, graph algorithms and data security, she has also contributed to research being undertaken by the Faculty's Centre for Interdisciplinary Built Environment Research which focuses on the Building Information Modelling environment that facilitates collaboration and information sharing but also introduces new threats to the security of the data it harbours.

A 2012 study into measuring attitudes towards privacy, medical research and consent by Professor Brankovic and her PhD student and colleagues found that while individuals greatly supported medical research, they also had concerns about the privacy of health information, and believed they should be asked for permission before their health information is used for any purpose other than medical treatment. However, the study also revealed that many people are not aware that simply removing their names and other direct identifiers from medical records does not guarantee their health information is protected.

As our awareness of privacy issues increases, so does the use of knowledge discovery and data mining (KDDM) techniques that are concerned primarily with discovering and analysing patterns in data. An interdisciplinary technique, KDDM analysis is underpinned by statistics, machine intelligence, pattern recognition, databases, optimisation, information visualisation and high-performance computing. It is a bit of a 'chicken and egg' situation: as computerisation and networking expand, so does the collection of massive amounts of data; as data grows in scope and size, there is a greater demand for KDDM technologies to suit these 'big data'.

Professor Brankovic, together with her colleague Professor Vladimir Estivill-Castro, was one of the pioneers of privacy preserving data mining in Australia who warned that during the process of data mining from data collection to knowledge discovery, data are exposed to several parties leading to a potential breach of individual privacy. Over the years, together with a team of her PhD students, she has, on one hand, developed various 'noise' addition techniques that protect privacy in data mining while preserving the patterns found in data. On the other hand, she has designed a suite of techniques that carefully restrict access to data to protect privacy while maximising the usability of data. These techniques are based on solid mathematical foundations and have performance guarantees.

In the area of algorithms, Professor Brankovic is collaborating with her German colleague Professor Henning Fernau to develop parameterised approximation algorithms for computationally hard problems, with the addition of privacy. This research is funded by the German Research Foundation and over the next two years Professor Brankovic will spend some months at the University of Trier as a Mercator Fellow.