K-anonymization for Improved Data Utility of Transactional Data Stream

Abstract

With the tremendous amount of transactional data generated due to e-commerce activities in today’s world, knowledge discovery from that data has become an es sential task for organisations to identify purchasing patterns of the users. But in the process of extracting useful information from the data the privacy of the users whose data is being used need to be kept safe. k-anonymization is one of the techniques to achieve this. Most of the work done in this area is on structured data, problem with transaction data is that it involves large number of different items and gener alization techniques that can be applied in case of structured data can’t be applied here. Some work has been done on anonymization of non-stream transaction data in which restructuring based techniques are used to keep similar transactions together. A method to apply k-anonymization on transaction data stream is proposed in this research work. A combination of sliding window mechanism and nearest neighbour algorithm is used to form a k-anonymous group.