Just take the aggregate of papers published on those four sites? That's certainly a rather crude methodology :p.