Abstract

Online reviews have become very popular these days. Every single thing that we buy is dependent on the reviews. Previously, people used to review movies only. Now, with the evolution of online shopping, people started reviewing electronic gadgets (phones, earphones, speakers, etc.), clothes, accessories, home furniture, restaurants and the list keeps going on and on. With the evolution of online reviews, the retailers have taken advantage of these and started paying the reviewers money to either give a positive review of their product or a negative review of their competitors product. This has led to the growth of counterfeit reviews. People who completely rely on reviews, blindly believe them and buy their products. To avoid this, some researchers have come up with a machine learning software called fake reviewer system. This fake reviewer system, eliminates all the reviews that are counterfeit and only displays reviews that are genuine to the people. But, then are these accurate? The answer is no. In this paper, we are proposing a fake reviewer system that uses Hadoop principles like wordcount and Hive query Language and couples it with supervised machine learning algorithms to achieve a better accuracy.