In some cases, simple functions such as the WAM will be adequate to meet the
goals of recommendation, with potential improvements to the RS lying in other
directions. Due to its properties, the WAM is quite a robust and versatile function. It
is not biased towards high or low scores, it does not accumulate the effects of errors,
it is computationally inexpensive and its common use makes it well understood
and easily interpreted. We present the power mean and Choquet integral as two
example alternatives whose properties might make them more appropriate in certain
situations.