Evaluation of Data Efficacy in Privacy-aware Computing

The Digital era marked by the unparalleled growth of Internet and its services with day-to-day technological advancements has paved way for a data driven society. This digital explosion offers huge market value to organizations and business processes who gain valuable information from the collected data to promote futuristic decision making. The aftermath of this large scale analytics is the privacy breach of sensitive information of individuals. The confidential information of people is at a risk of disclosure when it is used for a purpose not intended. Not only should privacy techniques guard classified data, but also ensure that it does not weaken data value when considered for analytical purposes. The proposed work focuses on assessing the expediency of sanitized data for analytics, using a threshold based optimal strategy for improving anonymization with large data. Classification models are trained on the anonymized data using benchmark algorithms and its accuracy is determined. The work is extended to evaluate the performance aspects of the proposed technique on diverse parallel execution environments, Hadoop and Spark.