Date of Award

May 2021

Degree Type


Degree Name

Doctor of Philosophy


Management Science

First Advisor

Huimin Zhao

Committee Members

Huimin Zhao, Yang Wang, Sanjoy Ghose, Cheng Chen


Business Analytics, Data Analytics, Data Mining, Machine Learning, Predictive Analysis, Text Mining


As online reviews become increasingly prevalent, both online businesses and customers face big data challenges. Individuals are now relying on reviews derived from websites where the reliability of a source depends on the reviewers. Customers spend much time and effort looking for reviews that are useful for them. Accordingly, online review platforms aim to explore various approaches to select useful reviews and present them to customers. At the same time, for business owners, marketers, and e-commerce managers, it has become an essential strategy in recent years to collect as many online reviews as possible. If marketers and managers are able to predict which customers would generate e-WOM (electronic word of mouth) content in the online community, they can come up with a practically effective marketing strategy. We explore online reviews from these two perspectives in the two essays of this dissertation.Essay 1 examines how to predict the most attractive reviews for a specific business entity. Previous studies have developed various methods to predict the helpfulness of online reviews. These methods have disregarded the aspects of the business entities when dealing with datasets for prediction and evaluation. They have not considered interactions between a review and the target business entity. This study proposes a novel method to predict the top attractive reviews for a specific business entity. We also suggest topic-related features to characterize the topics in a review and interaction features to reflect relationships between a review and the business entity it covers. Our empirical evaluation shows the utility of our proposed method and features. Essay 2 explores how to predict potential customers who are likely to write online reviews for a specific business. Marketers or e-commerce managers focus on finding individuals who can be deemed target customers and employ various techniques to gain a target market. One of the most common ways is providing promotional services to unspecified individuals. In this circumstance, many customers may consume just once to use the promotion out of the marketers’ expectation. As such, it is necessary to ensure that marketers have identified the target individuals who are prone to writing reviews of their consumption on online platforms. Business owners could benefit if they are able to predict potential customers who would generate e-WOM content for them in the online community. Then, the owners would provide valuable promotional services where it would be an efficient method to promote their online popularity while using minimal expense in the process. This research analyzes existing online reviews as examples of e-WOM using various features that reflect relationships between a business and a customer. In previous studies, researchers have relied on survey analysis to predict target customers who have the intention of generating e-WOM. However, this form of research can be distorted and thus faces issues when coming up with predictions for real businesses. Therefore, actual datasets are used in the current study to predict individuals who would write online reviews for a particular business. This research attempts, for the first time, to predict potential customers who would generate e-WOM and evaluate the prediction performance using actual online review data.