An Analysis of EWOM Text that Contribute to EWOM Helpfulness

Electronic markets are increasingly becoming first choice for shopping by customers. The electronic word of mouth (EWOM) plays a significant role in their online buying behavior and loyalty formations. It supports in reducing uncertainty and risk confronted by customers in online shopping. The EWOM helpfulness has emerged as a popular area of research both for business and academic communities in recent times. The role of comprehensibility and credibility of EWOM text and their impact on EWOM helpfulness has not been investigated much in existing research studies. In this paper, we build a EWOM helpfulness predictive model through machine learning (ML) methods and examine the influence of comprehensibility and credibility features implanted in the textual content of EWOM on the EWOM helpfulness. We have used text analytics and ML techniques to extract comprehensibility and credibility features from the EWOM text. In addition, linguistic and sentimental features of EWOM text have also been used for EWOM helpfulness prediction. A number of popular ML methods have been applied to predict the helpfulness of EWOM dataset extracted from Amazon.com. This study finds that comprehensibility and sentimental information of review text play an important role to outline EWOM helpfulness. Moreover, the number of right spell, wrong spell, adjectives, nouns, and adverbs embedded in EWOM text are also effective predictor of EWOM helpfulness.