Asymptotically Optimal Bandits under Weighted Information