Mean Percentile Ranking (MPR) explanation

问题

I am trying to use MPR as a metric to evaluate my recommendation system based on implicit feedback. Can somebody please explain MPR? I have gone through this paper

However, I can't seem to get an intuitive understanding.

Any help would be appreciated.

EDIT : I went through Microsoft's research on metrics for recommendation engine metrics

It is recommended that MPR is recommended when we're looking for one 'positive' result. Can somebody also explain why that is the case?

EDIT 2 :

来源：https://stackoverflow.com/questions/36720723/mean-percentile-ranking-mpr-explanation