For a review with helpfulness 2/3 (2 out of 3 people voted it helpful), we consider a numerical measure of its helpfulness as 2/(2+3) = 2/5 = 0.4. This normalizes the helpfulness scores to be predicted by participating systems. For reviews with helpfulness score 0/0, our normalized value is 0.
Participating systems will be evaluated on the basis of the closeness of their predicted scores (normalized) to the true scores (normalized).
Please refer to this document describing several measures of evaluation (as used in RevOpiD-2017 as well) for the evaluation of review rankings.