On data reduction for Boolean matrix factorization

In the associated paper, "On data reduction for Boolean matrix factorization,'' we examine existing work on data reduction for the problem of factorization of Boolean data and propose a new data-reduction method. This supplement to the paper contains two folders:

Source
This folder contains all benchmark datasets used for experimental evaluation.

Experimental results
This folder contains detailed results of experiments as described in the paper, i.e., comparisons of the included reduction methods for the reduction corresponding to selection of 50%, 25% and 10% of the input data; all experiments are evaluated using two benchmark algorithms, GreConD and Asso. The results are presented in two formats:
1) Supplementary Table X.pdf - This file contains a table reporting a maximum reached coverage, expressed as the average over 20 independent repeats, on the benchmark datasets for the studied reduction levels.
2) Charts - Files in this folder display the cumulative coverage of data preprocessed by each reduction method.