Publishing High-Dimensional Micro Data Using Anonymization Technique

Now a day’s society is experiencing very good growth in the count and variety of data collections having person-specific information as network connectivity, computer technology & disk storage space become increasingly affordable. Large databases is in use today’s society. The large amount of data available means that it is helpful to learn lot of individual information from public data. While doing, the privacy of the data should be persevered. The personal data may be misused, for various purposes. In order to improve these concerns, a number of techniques have newly been proposed to do the data mining tasks in a privacypreserving manner. So, to preserve the privacy of the data, privacy preserving methods are introduced and one of the methods is ‘anonymization’, in which a record is anonymized by using different anonymization techniques. Privacy preserving in data publishing is most important research area in data security area. Privacy-preserving data publishing provides methods and tools for sharing useful information while preserving data privacy and also analyzed many methods used for data privacy and their reward and short comes are monitored very well. Many techniques have been designed for micro data publishing with privacy preserving, such as generalization and bucketization. Several works resulted that generalization loses some amount of information specifically for high dimensional data. So it’s not efficient for high dimensional data for privacy preserving microdata publishing. In Bucketization, it does not prevents membership exposes and also does not suitable for data that do not have a clear separation between Quasi-Identifying attributes and Sensitive attributes of data. In our paper, an efficient method for data anonymization known as slicing and k-anonymity is introduced in which data can be partitioned both vertically(column) and horizontally(row). Advantage of slicing is that it works on high-dimensional data. Also, slicing preserves better data utility and privacy than any other method.An efficient algorithm is developed for computing sliced data that follow l-diversity requirement. A random permutation is done to randomly order a set of objects, that is, to permutate the data forapplying more privacy. Data slicing gives better utility than generalization and also no necessary for clear separation between Quasi-identifying and sensitive attributes. Experimental results using hospital dataaset confirm that data slicing provides data utility than generalization and more effectual than bucketization including sensitive attributes. Experiments we use the hospital patient datasets propose that our approach achieves better utility and also efficiency than any other existing and baseline algorithms while satisfying of proposed security work.