Do you mean it just indexes the short "abstract" that's provided of each paper? Or does this even go less, and just builds an index of the URLs only.