Research On Privacy Protection Of Transactional Data Stream Released

For decades,with the rapid development of information technology and application in daily life,the Internet-based application,large-scale data has generated.These data contains a large amount of personal information,which can service for business decisions or scientific research,makes data sharing becomes more and more important.However,these data containing personal sensitive information may leakage users privacy without suitable masking.Therefore,the research of data publishing privacy protection becomes the issue of current research.The main idea of which is to protect individual's privacy with less information loss.At present,most of the privacy protection techniques are designed for static data sets.With the promotion of internet of things and the coming of big data era,click logs,call records,and supermarket transaction data,are in the form of dynamic data stream,which is massive,real-time and dynamic changes.Traditional privacy preserving techniques are incapable for data stream.Transactional data streams is a typical kind of data streams where each record is a collection of items.This data usually contains sensitive information of users.The implemented privacy protection methods in data streams distribution are mainly for relational data streams.Privacy protection technology used to mask relational data streams cannot be directly applied to transactional data streams,which are high dimensional,sparse and dynamic.This paper studies the privacy protection for transactional data streams,and proposes a sliding window-based privacy protection method.The main research works are as follows.Firstly,we proposed the sliding window-based ?-uncertainty privacy protection model,which requires any sliding window satisfying ?-uncertainty.In addition,the effect of deleting sub-window and adding sub-window is analyzed by the information loss measure method for sliding window when,causing by the new coming data and deleted historical data,the sliding window is no longer meet the privacy requirements.Secondly,sensitive rule tree affected by adding or deleting sub-window is adopted to quickly find out the broken privacy requirement rules in the current window and,until the current window satisfies ?-uncertainty,suppress items as few as possible.To further reduce the information loss,we propose a method combining suppression and generalization,and,according to the information loss measure,judging whether to delete or generalize the item for making the current window meet the privacy requirement.Finally,the design scheme of the system and the detailed implementation of each module are given.From the two aspects of data anonymity efficiency and data utility,the proposed method is compared with the method by anonymizing a static transaction data.The experimental results show that the proposed method can anonymize data quickly and preserve the data utility effectively.