Find a way to separate high-quality and low-quality bibliographic entries

[JonathanReeve]

The tools that I wrote extract lots of URLs from syllabi, but they aren't doing a great job at guessing which URLs point to papers, articles, and other useful things, and so we're getting lots of readings that are actually just links to something like facebook.com/like-button or equivalent.

The Zotero translator does an OK job at trying to figure out which ones are legit, but we still need a way to separate high from low-quality bibliographic entries. This sounds like a job for a document categorizer?