Privacy Preserving Data Mining

A Comparative Study on Privacy Preserving Data Mining Techniques

Raza Hassan

Student, Dept. of CS

Nation University of Computer &

Emerging Sciences, CFD

[e-mail] Sajid Anwer

Lecturer, Dept. of CS

Nation University of Computer &

Emerging Sciences, CFD [e-mail] Abstract

Privacy preserving has devised as a vital concern with reference to the success of the data mining. Privacy preserving data mining (PPDM) deals with hiding an individual’s sensitive identity without sacrificing the usability of data and is a well-researched branch even today. People have become well conscious of the privacy intrusions on their personal data and are very hesitant to share their sensitive information. The data mining process if carried

On involving privacy and security with data mining, the privacy implies keeps useful knowledge about the discrete from being available to others[4]. So, there might be a clash between data mining and privacy. But, in reality, there is no need to interrupt privacy for data mining. Privacy is not disrupted till one feels his personal information is being used destructively. Once personal information is exposed, one cannot prevent it from being altered. For example D_O_B, father name and the gender etc. cannot become a risk for a specific, but if one more attribute like the distinctive ID or constituent ID are also known then it may source of a stern conclusion like identity theft[5].

In this paper, i discuss different efforts made in the area of Privacy of preservation of data mining (PPDM). This paper is organized as follows. In section II, i describe a literature review of privacy preserving data mining. In section III, i describe the framework of privacy preservation of data mining. In section V, i describe the application of PPDM. In section V, i discuss different challenges of PPDM. In section VI, I describe the different techniques or approaches for the PPDM. In section VII, I discuss the evaluation criteria of PPDM. Finally, i conclude in section

In “Deivanai, et al. [6]” the author use the hybrid approach for the Privacy preservation of data mining.In hybrid approach a combination of different techniques are used which combine to give an integrated result. For accuracy it uses anonymization and sppression to preserve data.In “Lohiya and Ragha [7]” the author use the hybrid approch for the privacy preservation of data mining.In this hybrid approch a combination of K-Anonymity and Randomization approch is used for privacy. It has a better accuracy and original data can be reconstructed.

In “Mathew and Obradovic [8]” the author use the decision tree for the privacy preservation of data mining.An approach which is technical methodological and should give judgmental knowladge.A graph based framework for preserving patiet,s sensitive information.In “Komishani and Abadi [9]” the author use the Trajectory data for privacy preservation of data mining.The approch is based on trajectory data publishing in which trajectories and sensitive attributes are generalized with respect to different privacy requirments of moving objects and it also to resistant to all three identity linkage and similarity