Hybrid recommendation strategies are often adopted to alleviate the cold-start problem, and nutritional goals (such as fat loss or muscle gain) can be incorporated into ranking models as constraints.