Uncertain Reasoning Using Maximum Entropy Inference