How Does Language Reactor rank words by frequency?

The title is pretty self explanatory. I was curious about what source material Language Reactor uses and how it creates frequency lists for each language.