Publicsdistinct

Publicsdistinct is a term used in data privacy and information theory to describe the degree to which records are distinguishable by attributes that are publicly observable. It serves as a metric for assessing how identifiable individuals may be when only public or linkable attributes are considered.

Formally, let D be a dataset and Q a set of public quasi-identifiers (attributes such as demographics or other publicly accessible markers). The publicsdistinct of D with respect to Q is the proportion of records in D whose Q-profile is unique, meaning there is no other record in D sharing the same values on all attributes in Q. A cross-dataset variant considers two datasets and evaluates how often a record in one dataset has a unique public profile when compared to the other dataset or to external public sources. These formulations reflect the idea that higher distinctness implies a greater potential to re-identify individuals based on publicly available information.

Usage and interpretation: higher publicsdistinct indicates that more records have unique public attribute combinations, which can correspond to higher re-identification risk if those publicly observable attributes are exposed or linked with other data. It is intended as a complement to established privacy measures such as k-anonymity, l-diversity, and t-closeness, focusing specifically on the impact of public information. It is sensitive to how attributes are discretized and to the external data landscape.

Limitations: publicsdistinct depends on the available public attributes and the external data environment, which can change over time. It does not directly quantify the protection of non-public attributes or account for all disclosure risks, but it provides a focused lens for evaluating privacy risk tied to public information.