System and method for local thresholding of re-identification risk measurement and mitigation
System and method to produce an anonymized electronic data product having an individually-determined threshold of re-identification risk, and adjusting re-identification risk measurement parameters based on individual characteristics such as geographic location, in order to provide an anonymized electronic data product having a sensitivity-based reduced risk of re-identification.
1. A system for local thresholding of re-identification risk, the system comprising:
a processor configured to:
quantify, based on variable security controls, contextual factors of the re-identification risk for different datasets comprising data representing each of a plurality of individuals, each individual having a plurality of fields;
for each individual in each dataset:
determine, based on sensitivity of the data and the threshold factors,
a local threshold of the re-identification risk specific to each individual,
a re-identification risk value of each individual, and
that the re-identification risk value is below the respective local threshold;
aggregate the re-identification risk value of plurality of individuals in the dataset, wherein the local threshold and re-identification risks are normalized across the plurality of individuals;
revise the local threshold based on respective data perturbation for the individual; and
place the individuals in one of a plurality of groups in accordance with the respective local threshold.
2. The computing system of claim 1 , wherein the processor is further configured to determine the sensitivity is using a questionnaire-based scoring matrix.
3. The computing system of claim 2 , wherein the sensitivity is specific to a region of the individual.
4. The computing system of claim 1 , wherein a classification of the plurality of fields depends upon inputs of the plurality of individuals.
5. The computing system of claim 1 , wherein the processor is further configured to determine the threshold factors for the dataset based on global sensitivity of the data.
6. The computing system of claim 1 , wherein the re-identification risk for each individual is based on a locally-computed regional inputs of the plurality of individuals.
7. A method for local thresholding of re-identification, the method comprising:
quantify, based on variable security controls, contextual factors of the re-identification risk for different datasets comprising data representing each of a plurality of individuals, each individual having a plurality of fields;
for each individual in each dataset:
determining, based on sensitivity of the data and the threshold factors,
a local threshold of the re-identification risk specific to each individual,
a re-identification risk value of each individual, and
that the re-identification risk value is below the respective local threshold;
aggregating the re-identification risk value of plurality of individuals in the dataset, wherein the local threshold and re-identification risks are normalized across the plurality of individuals:
revising the local threshold based upon respective data perturbation for the individual; and
placing the individuals in one of a plurality of groups in accordance with the respective local threshold.
8. The method of claim 7 , further comprising a step of determining the sensitivity using a questionnaire-based scoring matrix.
9. The method of claim 7 , wherein the sensitivity is specific to a region of the individual.
10. The method of claim 7 , wherein a classification of the plurality of fields depends upon inputs of the individuals.
11. The method of claim 7 , further comprising a step of determining the threshold factors for the dataset based on global sensitivity of the data.
12. The method of claim 7 , wherein the re-identification risk for each individual is based on a locally-computed regional inputs of the plurality of individuals.