A Visual Tool for Interactive Privacy Analysis and Preservation on Order-Dynamic Tabular Data

Anonymous

Keywords: visualization, privacy preserving, datacube, interactive systems

TL;DR: We propose a visual tool, TPA, for privacy protection of tabular data.

Abstract: The practice of releasing individual data, usually in tabular form, is obligated to prevent privacy leakage. With rendered privacy risks, visualization techniques have greatly prompted the user-friendly data sanitization process. Yet, we point out, for the first time, the attribute order (i.e., schema) of tabular data inherently determines the risk situation and the output utility, while is ignored in previous efforts. To mitigate this gap, this work proposes the design and pipeline of a visual tool (TPA) for nuanced privacy analysis and preservation on order-dynamic tabular data. By adapting data cube structure as the flexible backbone, TPA manages to support real-time risk analysis in response to attribute order adjustment. Novel visual designs, i.e., data abstract, risk tree, integrated privacy enhancement, are developed to explore data correlations and acquire privacy awareness. We demonstrate TPA's effectiveness with two case studies on the prototype and qualitatively discuss the pros and cons with domain experts for future improvement.