For a resource that helps identify predatory journals, try Beall's list (ironically the source of the article ;o).
As an academic, I don't think Google Scholar should filter out papers from predatory journals because in science the source of an argument is irrelevant, what matters is the correctness of the argument and the support it receives from the data. Peer review is only the first step in acceptance from the research community, not the last. The number of citations a paper has is a good indication of its acceptance by the research community and Google Scholar already sorts papers by citations. So google does filter by a measure of quality, just not in the way that Beall wants it to.
Just because a paper is published in a predatory open access journal does not mean it is a bad paper (although it is suggestive), so omitting papers from predatory journals will stop people from finding some pieces of good work that have been published there. Filtering by citations will hide most of the junk, but without hiding the good papers that have been published in not so good journals.
There is also the problem of the definition of "junk science" to consider. Some might say that acupuncture, or homeopathy, or the theory that the sun is made of iron, or climate change skepticism, or autism/vaccination links are "junk science". Whether this is true depends on the nature of the argument and the reasoning put forward, and you can't judge by keyword, you do have to look at the paper to find out. Good science quite often turns out to be wrong, so you can't judge what is junk science by whether it is right or wrong. What differentiates between good science and junk science has more to do with whether it follows scientific method (e.g. Popper's criterion of falsifiability).
If someone conducts a double blind trial to determine whether homeopathy actually works, then that is good science. If someone writes a paper on the mechanism underpinning homeopathy that is non-falsifiable, then that might arguably be junk science. By that definition, Google Scholar has almost no junk science, although it has plenty of bad science.
There is a good reason why it has plenty of bad science and scientific papers that are wrong, which is that science is performed by scientists, who are only human and hence fallible, and working at the very edges of what we know, where errors are much easier to make than on more mundane questions.
I don't find the article particularly persuasive. Google scholar actually does quite a good job of exposing bad science by demonstrating how little it is generally cited.