A survey on applications of opinion mining class imbalance data

Abstract

Opinion mining or sentiment analysis is to analyze the useful information from the large quantity of text messages or reviews regarding a product or a topic. In binary product reviews (positive or negative) the distribution of classes will always tend to any one class, thereby generating a class imbalance nature in the dataset. A class imbalance state of the dataset is in which, instances in one class predominately outnumber the instances in other class. The existing opinion mining approaches are not efficient on the class imbalance opinions mining datasets. In this paper, we present the up to date survey of class imbalance opinion mining datasets.