An Efficient Privacy Preserving Classification Tree Technique in K-Anonymity for Secure Data Mining and Data Publishing

In recent years of data mining applications, an effective technique to preserve privacy is to anonymize the dataset that include private information before being released for mining. Inorder to anonymize the dataset, manipulate its content so that the records adhere to k-anonymity. Two common manipulation techniques used to achieve k-anonymity of a dataset are generalization and suppression. However, generalization presents a major drawback as it requires a manually generated domain hierarchy taxonomy for every quasi identifier in the dataset on which k-anonymity has to be performed. In this paper, new method for achieving k-anonymity based on suppression is proposed. In this method, efficient multi-dimensional suppression is performed, i.e.,values are suppressed only on certain records depending on other attribute values, without the need for manually-produced domain hierarchy trees. Thus, this method identify attributes that have less influence on the classification of the data records and suppress them if needed in order to comply with k-anonymity. The method was evaluated on several datasets to evaluate its accuracy as compared to other k-anonymity based methods. Additionally, a new revised algorithm of kactus called ‘CombS’ can be used.