A nixified experiment to evaluate performances of different learning algorithms (more or less transparent) for multi-label classification problems.