Although reviews play a pivotal role in consumer decision-making, due to the large volume of reviews, consumers cannot read and evaluate all of them. To reduce the search and evaluation costs, online platforms can prioritize the most helpful reviews for consumers. Previous literature established a relationship between sentiment and helpfulness. Therefore, sentiments (i.e., positivity, negativity, and neutrality) that are automatically identified from review can be used to estimate helpfulness. Using 3,508 hotel reviews, we found that the non-linear Cobb-Douglas utility function best represents the relationship between review sentiments and review helpfulness, and it outperforms an OLS regression approach that has been the standard in the literature so far. Additionally, we found that text neutrality has the greatest impact on review helpfulness compared to other sentiments. Our findings can help to estimate review helpfulness when it is unavailable by utilizing a nonlinear utility function trained on data with known helpfulness values.