Privacy Preservation of e-Governance Data Using Local Beam Search
Manas Ranjan Patra, Debasis Mohapatra · 2018
E-governance is mandated to extend various digital services to the citizens in a transparent and easy-to-access manner. This requires an enormous amount of data acquisition, analysis, and processing. Normally, such databases contain references to private information of different stakeholders. It is the responsibility of the government to protect the data against privacy disclosure which can be maneuvered through direct or indirect inference by linking different databases. Though governments have adopted policies to control privacy but still there are technical issues and even the data analyst may turn out to be the one responsible for privacy disclosure. k-anonymity is one of the frequently used approaches to ensure privacy preservation in such scenarios by transforming a database to a sanitize database such that the probability of disclosure is minimized. However, this may lead to information loss and thus reduce the utility for data analysis. Therefore, it is always a challenge that while providing privacy of data through k-anonymity one must ensure minimum information loss without hampering the utility of data. Towards this end, we analyze the efficiency of two existing greedy heuristic based techniques, viz., Datafly and Improved Greedy heuristic and we propose two Local Beam Search based algorithms: Sequential Local Beam Search (SLBS) and Binary search based Sequential Local Beam Search (BSLBS). A comparative analysis of the algorithms is done with the help of Information Loss metric Info.Loss. It is observed that BSLBS outperforms the existing greedy based heuristic methods.