SPARK-Based Partitioning Algorithm for k-Anonymization of Large RDFs

Privacy protection for resource description framework data is very important because RDF (i.e., linked data) is widely used in published data format in many areas, including government open data, health-care for individuals, and social relationships. As data can include private information belonging to individuals or companies and can make private information available to third parties, there are several anonymization models provided for preserving privacy in practice. k-anonymity has thus gained attention in research. Recently, several RDF anonymization models have been proposed. However, current approaches focus on a model and a metric for measuring information loss but do not consider large-scale RDF data. In this paper, we propose an efficient anonymizing method for large-scale RDF data. We develop a greedy partitioning algorithm (i.e., SPARK) for RDF anonymization. SPARK is a leading platform for big data processing. The results of experiments on synthetic datasets demonstrate that our proposed method requires less running time than previous methods.