Analysis and prediction of extent of helpfulness of reviews on E-commerce websites

Abstract

Online reviews on e-commerce websites play quite a decisive role in customers’ purchase decisions on these platforms. The current ranking patterns of these platforms are mostly based on the ratio of helpful and total votes. Thus if a recent review which is also helpful cannot get a good ranking due to this ranking pattern. With an increasing number of reviews on these e-commerce websites, even an average product can get enough reviews making it a difficult task for consumers to go for each review. Formerly researchers have mostly worked only on the helpfulness of review but very few have worked on the extent of review helpfulness. Our research examines the textual content of reviews on e-commerce websites with different helpfulness votes to further classify a new review and give the recently submitted review a proper rank or place in the current set of reviews i.e. our model can not only decide helpfulness of review but can also decide up to what extent a particular review is helpful by applying different machine learning algorithms. Thus an e-commerce website consumer could be highly benefited from this as the consumer doesn't have to go through all the reviews before purchasing.