The Research And Application Of Privacy Preserving Based On Granular Computing

It has become much easier to collect and use information about individuals via internet. In term of unauthorized users, the publicity of individual information will cause the privacy problem, even if not intentionally. Of course, if used appropriately, these data may be very valuable sources for scientists, analysts, and policy-makers. So, if there is no effective measure, there may be a great danger of privacy invasion. Therefore, how can we balancing data utility and privacy protection effectively is the core problem for the process of implication.Granular Computing ("GrC" for short) is the new concept and computing model to information proceeding, and then the main idea is problem solving on different granular hierarchies. In the paper, granular computing is applied to the field of privacy protecting based on studying the concept of granular computing deeply, so that it is aimed at seeking for a privacy protecting method based on Granular Computing.In the paper, the important research is study and application of privacy protecting based on rough sets and granular computing model in connection with incomplete information system. Firstly, based on the study of GrC concepts and models, a method of how to construct hierarchical rolerance granularity space based on reduct attributes is proposed, and design the correspond algorithm. According to the power sets of reduct attributes, granular knowledge of each hierarchy is constructed, and is ready for follow algorithms. Secondly, a method based on GrC and algorithms of information anonymisation is proposed. The method is viewed as a measurement of approximation classified quality over decision information, which test information anonymisation whether or not. Then this method is achieved from follow steps: First, the reduction of the original information and approximation classified quality of the reduct attributes over decision information is obtained. Second, if classified quality of the original information system the approximation is not satisfied, hierarchical compatible granularity space of incomplete information system is constructed. Third, the hierarchical tolerance granularity space of original information system is traversed and made the correspond attribute values coarser: from first hierarchy, attribute values of all the attributes, that is the set of reduction attributes, are coarsened on this hierarchy, until the approximation classified quality of the coarsened information system over the set of reduction attributes on this hierarchy is smaller than that of the original information system over the set of reduction attributes. Finally, in order to prove the efficiency of the method of information anonymisation and the correspond algorithm, three groups of testing datasets are chose to test from various aspects, and analysis the testing results. Then the testing results show that the method of information anonymisation based on GrC is effective.In last part of the paper, all of researches and work in the paper are summarized, and many aspects of improvements are analyzed, and the further research directions in the field of privacy protecting based on rough set and granular computing are prospected.