In your role at a major social network, you're tasked with improving the efficiency of human moderators by developing a Machine Learning model to flag potentially offensive comments for review. The platform processes millions of comments daily, and the model must balance between identifying as many offensive comments as possible (high recall) and minimizing the number of non-offensive comments incorrectly flagged (high precision). Given the constraints of processing power and the need for timely moderation, which metric(s) would best evaluate the model's effectiveness? Choose the two most appropriate options.