Sentiment Analysis on Reviews: Understanding eWOM Using Deep Learning

Word-of-mouth (WOM) has changed ways of marketing communication between marketers and consumers, as well as among consumers. The development of information technology and e-business has saliently facilitated the generation and spread of eWOM. Given millions of consumer reviews, however, limitations of traditional analytical tools hinder a systematic understanding of reviews as big data and their business implications. In this study, we developed methods basing on machine deep learning to analyze massive online reviews. In particular, we systematically quantized reviews to sentiments, i.e. evaluative effective responses. We found that the techniques from deep learning provide a high accuracy quantization from reviews to evaluative effective responses. Theoretical and practical implications are discussed.