What do they mean by entropy of a string? Entropy of the distribution of frequencies of characters? N-grams?

In this particular case, we're referencing Shannon Entropy. We've got a few other items that also look at bigrams on a per-language basis. As you can imagine, this changes depending on the language used for the given software package.

The table of typical entropy of those encodings is extra useful if you hack around with CyberChef. Embedded and nested zip files would be another common encoding mechanism used to bypass AV.

"Large numbers of obfuscated strings are uncommon in benign software. "

I think this is simply not true when it comes to the web.

So the next obfuscation technique would be to encode strings using a low entropy encoding, yeah?