A note on belief structures and S-approximation spaces