Sort:  

You mean it also uses machine learning to analyse the language of the text? Damn!
I thought I was the first to come up with this idea...

I thought others would only analyze metadata, and not the text itself.

I don't recall the technical details, but it's on GitHub