robinlin/dptable repository overview

Run command

Rscript exp/run-category.R -f <DATA_NAME> -e1 <EPSILON_1> -e2 <EPSILON_2> -nrun <RUN_TIMES>

-f: dataset name -e1: epsilon value for constructing noisy junction tree -e2: epsilon value for injecting noise into the marginal tables -nrun: num of runs, 10 as default -CV: numeric value in (0, 1), threshold Cramer’s V value for picking correlated attributes pairs, 0.2 as default,typically choose 0.2 for weakly correlated datasets; 0.3 for highly correlated datasets -q: boolean value indicates whether to process query on the releasing tables, True as default -sim: boolean value indicates whether to simulate and generate synthetic data, False as default. If setting as True, the program shall generate one copy of synthetic data with raw data's size in the output directory.