Predicting the sentiment of SaaS online reviews using supervised machine learning techniques

Abstract

There has been a dramatic increase in the sharing of opinions and information across different web platforms and social media, especially online product reviews. Cloud web portals, such as getApp.com, were designed to amalgamate cloud service information and to also examine how consumers evaluate their experience of using cloud computing products. The current literature shows the growing importance of online users' reviews, hence this study focuses on investigating consumers' feedback on Software-as-a- Service (SaaS) products by developing models to predict reviewers' attitudes. The goal of this paper is to develop prediction models to predict the sentiment of SaaS consumers' reviews (positive or negative). This research proposes five models that are based on five algorithms, the Support Vector Machine algorithm, Naive Bayes algorithm, Naive Bayes (Kernel) algorithm, k-nearest neighbors algorithm, and the decision tree algorithm to predict the attitude of SaaS reviews. The prediction accuracy of the space vector algorithm (5-fold cross-validation) is 92.37% which suggests that this algorithm is able to better determine the sentiment of online reviews compared with the other models. The results of this study provide valuable insight into online SaaS reviews and will assist in the design of SaaS review websites.