Abstract:
Online user reviews are quite popular in social media, e-commerce and review websites. It is commonly referred as word of mouth which provides positive and negative messages from users about products and services. It helps users to get insights through review ratings and subjective feedback. As the volumes of reviews are high, it makes it harder for users to identify the helpfulness upfront. In general helpfulness rating is provided by the users who read the review, but many reviews still stay unrated. In this paper, we propose an approach of predicting helpfulness of such reviews from mouthshut.com using a combinatorial approach of empirical analysis and naïve Bayes machine learning method. The data set is chosen for Indian Online Travel Agencies (OTA) namely Makemytrip, Cleartrip, Yatra, Goibibo, and Expedia India. A detailed experiment is conducted and results are discussed by analyzing review metadata characteristics.