OptimalAlgorithmsforStochasticContextual PreferenceBandits