The Impact Factors of Online Book Reviews Usefulness: An Empirical Comparison Between ANN and SVM

The rapid development of information technology has led to massive online reviews which are generated on the Web, this paper aims to explore the determinants of review helpfulness. In this research, we propose a conceptual model from the perspective of content quality and source quality and incorporate supervised machine learning approaches to learn the impact factors of online book reviews. Specifically, we compared two supervised machine learning algorithms of ANN (Artificial Neural Network) and SVM (Support Vector Machine) and Linear Regression approach. Book reviews are collected from douban.com for ten popular books. We found that long sentences and the extreme emotion is more helpful and can be more persuasive. The empirical findings also indicated that the ANN and SVM algorithms outperformed the ordinary OLS algorithms.