Exploring the Security Vulnerability in Frequency-Hiding Order-Preserving Encryption

Table 3

Rules resulting from the Apriori algorithm applied to the Hospital Inpatient Discharges (SPARCS De-identified) dataset.

No	Antecedents	Consequents	Antecedent support	Consequent support	Support	Confidence	Lift	Leverage	Conviction
1	(653)	(70 or Older)	0.025343	0.186906	0.016895	0.666667	3.566855	0.012159	2.439282
2	(109, M)	(50 to 69)	0.013200	0.359556	0.010032	0.760000	2.113715	0.005286	2.668515
3	(109)	(50 to 69)	0.021119	0.359556	0.012672	0.600000	1.668722	0.005078	1.601109
4	(233)	(M)	0.022770	0.167932	0.012144	0.884615	1.561474	0.004367	3.756776
5	(5)	(M)	0.026927	0.566526	0.019007	0.705882	1.245984	0.003752	1.473812
⋮	⋮	⋮	⋮	⋮	⋮	⋮	⋮	⋮	⋮

Each column is explained as follows: “Antecedents” are the items that precede, and “Consequents” are the items that follow. “Antecedent Support” and “Consequent Support” show the proportion of transactions in the data that contain the antecedent and consequent, respectively. “Support” indicates the frequency of the antecedent and consequent appearing together, while “Confidence” shows the conditional probability of the consequent given the antecedent. “Lift” measures how much more likely the antecedent and consequent are to occur together than if they were statistically independent, “Leverage” computes the difference between the observed frequency of the antecedent and consequent appearing together and what would be expected if they were independent, and “Conviction” indicates the dependency of the consequent on the antecedent.