A Hybrid Imbalanced Data Learning Framework to Tackle Opinion Imbalance in Movie Reviews

Abstract

Opinion Mining is an important buzzword in recent times for research and industry for data science applications. Many concerns in opinion imbalance particularly in movie reviews were analyzed and handled for efficient recommendations. Opinion imbalance in terms of binary class, which often compromises the classifier prediction results, was scarcely studied. In this work, a Hybrid Imbalanced Data Learning Framework (HIDLF) is proposed to handle the opinion imbalance in the movie review dataset and then classify the movie reviews through the proposed HIDLT-SVM algorithm, which is a part of HIDLF for effective movie review classification. Experimental comparisons of the proposed work are done on movie reviews with Logistic Regression, CART, and REP Tree. Different evaluation metrics are used for capable classification of opinions from the movie reviews. The results recommend that the planned HIDLT-SVM framework performs better than the competent algorithms.