Online customer reviews have become a popular source of information that influences the purchasing decisions of many prospective customers. However, the rapidly increasing volume of online reviews presents a problem of information overload, which makes it difficult for customers to determine the quality of the reviews. This study defines the helpfulness of the reviews as a count variable and takes the review helpfulness prediction from both regression and classification perspectives. The influence of friends and followers on review helpfulness is examined by introducing Social Network Strength (SNS) features. Furthermore, the performance of Machine Learning (ML) algorithms and the importance of features are separately examined for both problems using different time span of reviews. The evaluation performed using a dataset of 90,671 Yelp shopping reviews demonstrates the effectiveness of the proposed approach. The findings of this study have important theoretical and practical implications for researchers, businesses, reviewers and review platforms.

在线客户评论已成为影响许多潜在客户购买决策的流行信息来源。但是，在线评论数量的快速增长带来了信息过载的问题，这使得客户难以确定评论的质量。这项研究将评论的有用性定义为计数变量，并从回归和分类的角度进行评论的有用性预测。通过介绍社交网络强度（SNS）功能来检查朋友和关注者对评论帮助的影响。此外，对于机器学习（ML）算法的性能和功能的重要性，使用不同的评论时间跨度分别针对这两个问题进行了检查。使用90个数据集进行的评估，671 Yelp购物评论证明了该方法的有效性。这项研究的发现对研究人员，企业，评论者和评论平台具有重要的理论和实践意义。