Abstract:Personally identifiable information (PII) can find its way into cyberspace through various channels, and many potential sources can leak such information. Data sharing (e.g. cross-agency data sharing) for machine learning and analytics is one of the important components in data science. However, due to privacy concerns, data should be enforced with strong privacy guarantees before sharing. Different privacy-preserving approaches were developed for privacy preserving data sharing; however, identifying the best privacy-preservation approach for the privacy-preservation of a certain dataset is still a challenge. Different parameters can influence the efficacy of the process, such as the characteristics of the input dataset, the strength of the privacy-preservation approach, and the expected level of utility of the resulting dataset (on the corresponding data mining application such as classification). This paper presents a framework named Privacy Preservation as aService (PPaaS) to reduce this complexity. The proposed method employs selective privacy preservation via data perturbation and looks at different dynamics that can influence the quality of the privacy preservation of a dataset. PPaaS includes pools of data perturbation methods, and for each application and the input dataset, PPaaS selects the most suitable data perturbation approach after rigorous evaluation. It enhances the usability of privacy-preserving methods within its pool; it is a generic platform that can be used to sanitize big data in a granular, application-specific manner by employing a suitable combination of diverse privacy-preserving algorithms to provide a proper balance between privacy and utility.

No One Size (PPM) Fits All: Towards Privacy in Stream Processing Systems

A Path to Holistic Privacy in Stream Processing Systems

Differential Privacy for Protecting Private Patterns in Data Streams

PrivacyStreams: Enabling Transparency in Personal Data Processing for Mobile Apps.

PrivStream: Enabling Privacy-Preserving Inferences on IoT Data Stream at the Edge

PPaaS: Privacy Preservation as a Service

PrivStream: A Privacy-Preserving Inference Framework on IoT Streaming Data at the Edge

PP-Stream: Toward High-Performance Privacy-Preserving Neural Network Inference via Distributed Stream Processing.

DPI: Ensuring Strict Differential Privacy for Infinite Data Streaming

Privacy-aware Process Performance Indicators: Framework and Release Mechanisms

Differentially Private Stream Processing at Scale

Streaming Data Collection With a Private Sketch-Based Protocol

Protection of Privacy in Pervasive Computing Environments.

Data-adaptive Privacy-Preserving Mechanism for Data Stream Pub-Lishing in Real-Time

P3LS: Partial Least Squares under Privacy Preservation

Online Context-aware Streaming Data Release with Sequence Information Privacy

PMDG: Privacy for Multi-Perspective Process Mining through Data Generalization

Anonymous Privacy Preservation Based on M-Signature and Fuzzy Processing for Real-Time Data Release

OptStream: Releasing Time Series Privately

DSPM: A Platform for Personal Data Share and Privacy Protect Based on Metadata

Extended Capabilities for a Privacy-Enhanced Participatory Sensing Infrastructure (PEPSI)