Private Frequent Item Set Mining in Smart Splitting for Privacy Preserving using Attribute Probability Matrix Saranya K.*, Premalatha K.** *Research Scholar, Computer Science and Engineering, Bannariamman Institute of Technology, Erode, India **Professor, Computer Science and Engineering, Bannariamman Institute of Technology, Erode, India Online published on 14 October, 2016. Abstract Privacy preservation becomes more stimulating task while publishing data which is maintained by any organization where there occurs isolated information about their customers. Data mining tasks are performed on published data for business intelligence or any of knowledge retrieval tasks. The task of sanitizing original data set should not loss the uniqueness of the data and should be able to restructuring the unique data from the sanitized one. This paper is to design a differentially isolated FIM algorithm based on the FP-growth algorithm. It consists of preprocessing phase and mining phase. The smart splitting method is used to limit the length of the transactions. But it gives some private information loss. To reduce the information loss we put forward a new ontology based sanitization technique which uses probability values of sensitive items to generate the data. Finally, the proposed method generates an ontology file to represent the publishing data from the probability matrix. Top Keywords Sanitized Data, Differential Privacy, Frequent Itemset Mining, Transaction Data, Privacy Preservation. Top |