I have a dataset where data can have multiple correct labels. If a label matches from the targeted list, it needs to be treated as accurate. What metric should I use in building an ML model with it?