You probably found this link from our 2019 paper. Unfortunately, we updated our site but the new website does not yet contain a page with a description of our dataset selection strategy.
You will be redirected to the markdown file from which the old page was generated, so you can still view the old text.
If you are not redirected within 5 seconds, you can find it at https://github.com/openml/automlbenchmark/blob/2fe3bd41768ce28387f827791bd57ef1a5a84783/docs/benchmark_datasets.md.