I work at a university. The lecturers have had software for years which checks for plagiarisation.
I've long been interested in just how such a software package works. How does it determine what is and is not a plagiarized work? Precisely what is measured?
It compares the essay to a web-search, other essays that have been searched previously, news papers and magazines that are provided electronically etc etc.
If I remember correctly, it provides you with a highlighted copy with links to the original material. It's quite good as filtering out quotes and the like.