Investigations on bio inspired feature selection techniques for spam review classification

Abstract

newline Online reviews are often contemplated as the primary factor in beholding customer s decision to purchase a product or service It is observed as a valuable source of information in determining public opinion of these products and services On considering this key aspect and their impact over the products manufacturers and retailers are highly concerned and are fretful about the customer feedback and reviews The increase in the reliance on online reviews has a direct effect on the sale of products since it may also induce the fake reviews to promote or devalue the products and their services This practice is known as Opinion Review Spam where the spammers manipulate and produce poisonous and fake reviews ie making fake untruthful or deceptive reviews to incur profit or gain In view of the fact that not all online reviews are truthful and trustworthy it is essential to develop techniques for detecting review spam By means of extracting meaningful features from the text using machine learning techniques for classification and evolutionary algorithms for feature selection In addition the reviewer information apart from the text can be used to aid in spam review classification process Moreover majority of current contemporary research has focused on supervised learning methods which requires labelled data It is a scarcity while dealing with the online review spam The web contains large possessions and assets of opinions about products politicians and more, which are expressed in different means including news group posts review sites The number of customer reviews received for a product is growing exponentially at a rapid rate