classification algorithms aim to predict an unknown label (e.g., a quality
class) for a new instance (e.g., a product). Therefore, training samples
(instances and labels) are used to deduct classification hypotheses. Often, it
is relatively easy to capture instances but the acquisition