Automatic detection of sensitive attribute in PPDM

Abstract

With rapid growth in technology, networking and reduced cost of storage media enabled the organizations to collect huge volume of information from heterogeneous sources. The various sources through which the information is collected are banking, healthcare system, insurance companies, government organizations etc. The information collected in this manner is treated as an asset with great quality, research value and is used for analysis purpose by organizations. Various data mining techniques are applied on such huge data to acquire useful [1] and relevant knowledge. Such information is not only used by organization but also released for public to acquire knowledge and benefits from such data. Sometimes dissemination of such data [2] which includes sensitive and personal information becomes the main source of sign of danger to violate the privacy of an individual. The simplest solution for this problem is to transform the sensitive information before disseminating it to the outside world. The data transformation should be performed in such a manner that the intruder should not be able to extract the true sensitive information and the released data should be valid for data analysis. Privacy of sensitive information can be protected by replacing the true data by false but realistic ones. Generally the database owner predefines the sensitive attributes whose data values are to be scrambled. The crucial task to recognize the sensitive attributes in the database has not been developed. In this paper we propose an innovative approach which accepts the user queries consisting of different attributes and identifies the sensitive attributes whose values are to be scrambled depending on threshold value. The threshold value is calculated depending on the different weights assigned to individual attributes. The information under those particular attributes whose total weights exceeds the threshold values is scrambled.