Research Of Hiding Sensitive Rules Algorithm

As an important branch of Privacy-protecting Data Mining, Sensitive Rules Hiding is highly valued by more and more researchers. The application of Sensitive Rules Hiding encompasses a wide range of fields, mainly in business competition, cooperation and finance. There are several limitations of the existing algorithms: many of them set the frequency of scanning the database according to the number of sensitive rules, which requires much time spending. On the other hand, many algorithms greatly affect the database when hiding the sensitive rules, reducing the availability of processed data sides. Therefore, it is necessary to develop a Sensitive Rules Hiding algorithm which reached to a balance point between protecting privacy and reducing the impact of the original data sides.Based on the study and analysis of the domestic and international technology of hiding sensitive rules, the article improves the defects by reducing the impact of the database and increasing the time efficiency. There are two main points of this article: first, to form the basis of the implementation of the improved algorithm, a model which scans the database twice is chosen and the structure as well as the operation of the intermediate files is improved. Second, aiming at the shortcomings of the existing algorithm which applies the data blocking method, an improved algorithm for hiding sensitive rules is proposed. There are three aspects of the algorithm: firstly, the conflict degree is regarded as the benchmark to select the sensitive matters, and there is overlap between the processing rules. Meanwhile, the repeated selecting of the same transaction and items are prevented to reduce the number of impact on the data sides. Second, when selecting the victim items, multiple items within the rule are selected and different support services are rotated respectively, one of them is selected and processed. Thirdly, by interacting with the intermediate files in the model, the database is completely processed after scanned twice. All the experimental results and theoretical analysis have verified the validity and accuracy of the algorithm and the improvement of the availability of the data sides.