Predicting the Helpfulness of Online Customer Reviews

Customers rely on online reviews when making purchase decisions on e-commerce platforms. Reading the massive amount of available reviews on products is a timeconsuming process. Hence, several online platforms allow customers to vote on whether a review is helpful. However, customers may consider the helpfulness voting process hard due to the number of reviews. This research paper presents an automated review helpfulness prediction method called HelpPrd. HelpPrd adopts text mining and machine learning techniques to predict the review helpfulness. The presented approach extracts features from the textual reviews to train three helpfulness classifiers. The experimental results demonstrate that HelpPrd-logistic regression achieved better performance than other implementations of HelpPrd in terms of accuracy, precision, recall, and F1-score.