“…The utility functions under consideration may be as simple as a (weighted) sum of criteria, or may be selected within a parameterized class of functions whose parameters are to be determined. This type of approach has been extensively investigated, in particular, by researchers interested in multiple criteria problems with discrete alternatives (MCDA) (see, e.g., , and in particular Bouyssou and Pirlot (2016); Dyer (2016); Moretti et al (2016); Siskos et al (2016) for recent surveys of closely related topics; see also Corrente et al (2016) for extensions), or in conjoint analysis (see, e.g., Giesen et al (2010); Gustafsson et al (2007);Rao (2014)). More recently, similar questions have also been investigated in preference learning, a subfield of artificial intelligence (see, e.g., Corrente et al (2013);Fürnkranz and Hüllermeier (2010)).…”