Could you publish the scripts you ran to obtain the dataset, so we can reproduce an up-to-date version?