Privacy-Aware Analysis based on Data Series

Stephan A. Fahrenkrog-Petersen, Han van der Aa, Matthias Weidlich

Abstract: Data that is recorded about the operations of an organization constitutes a valuable source of information for monitoring and improvement. Specific use cases include the assessment of compliance to legal regulations, the analysis of performance bottlenecks, or the optimization of resource utilization. In recent years, a plethora of algorithms for operational analysis using data series, summarized as process mining, have been developed to support these use cases, e.g., by constructing models for simulation and prediction or by comparing the recorded data against a normative specification of a process. Data series often contain sensitive information, though, about the individuals that act as service consumers or service providers. Personal information is only partially hidden by obfuscation and pseudonymization and potential privacy breaches need to be prevented for ethical, legal, and economic reasons. This tutorial is devoted to methods for privacy-aware analysis using data series. It covers essential notions, reviews privacy-disclosure attacks, and outlines techniques to give formal privacy guarantees while largely maintaining the data's utility for operational analysis. The discussion is structured by the adopted perspective on the privacy of individuals, and the degree to which a data series contains contextual information.