Exploiting Support Sets for Answer Set Programs with External Evaluations