Heuristic Approach on Protecting Sensitive Frequent Itemsets in Parallel Computing Environment
Dedi Gunawan, Guanling Lee · Kagoshima Kenritsu Tanki Daigaku Chiiki Kenkyūjo kenkyū nenpō · 2017
Due to the development of data mining algorithm, the process of extracting information can be done easily. However, in the opposite side, the sensitive knowledge may be revealed which causes privacy breach problem. Sensitive knowledge is the information of people and organization that should be kept in secret according to some privacy policy before it goes to public. Thus, Privacy Preserving Data mining (PPDM) is emerging and becoming the crucial part in data mining study. Therefore, in this work an efficient approach is proposed to avoid privacy breach in frequent itemsets mining. Owing to the huge amount of data and to boost the hiding process, we segregate the transactions into two parts where the first part contains all the sensitive transactions and the other part contains nonsensitive transactions. The first part is then split into several data chunks and sent to the servers. This step is followed by deciding which item that is going to be removed from transactions to perform data sanitization. A set of experiment is performed to show the benefit of our approach. According to the experimental results, the proposed approach works considerably well in hiding sensitive itemsets as well as keeping less changes in sanitized database.