Efficient search for the top-k probable nearest neighbors in uncertain databases