Data privacy employing a k-anonymity model with probabalistic match self-scoring

1. A computer-implemented method of protecting data comprising:

determining a desired duplication rate for a data set based on a level of desired anonymity for the data in the data set, wherein at least one data record in the data set includes fewer attributes than a corresponding original record and the desired duplication rate is (1−1/K), where K is an anonymity parameter;

generating a self-score threshold for data records within the data set based on the desired duplication rate and Receiver Operating Characteristic (ROC) curves including a threshold determined from a logarithm of N/(10+9K), where N is the number of data records within the data, wherein a matching function compares a first data record with a second data record to compute a matching score which is a logarithm of a ratio of a first value to a second value, wherein the first value represents a likelihood that the first data record and the second data record both relate to a same entity, and the second value represents a likelihood that the first data record and the second data record relate to different entities;

producing a data record self-score for each data record within the data set based on a comparison of attributes for that data record to attributes of the corresponding original data record, wherein the data record self-score for a data record is the result of applying the matching function to compare the data record to the corresponding original data record;

comparing the data record self-scores to the self-score threshold; and

controlling access to the data records of the data set based on the data record self-scores satisfying the self-score threshold.

2. The computer-implemented method of claim 1 , wherein controlling access to the data records includes:

enabling access to data records in response to the data record self-scores satisfying the self-score threshold indicating a presence of the desired level of anonymity for the data records.

3. The computer-implemented method of claim 2 , wherein the desired level of anonymity for the data records occurs when data record self-scores are less than or equal to the self-score threshold.

4. The computer-implemented method of claim 1 , wherein controlling access to the data records includes:

modifying one or more attributes of data records lacking the desired level of anonymity to provide the desired level of anonymity for those data records.

5. The computer-implemented method of claim 4 , wherein the lack of the desired level of anonymity for the data records occurs when data record self-scores are greater than the self-score threshold.