A Reinforced Deep Belief Network with Firefly Optimization for Sentiment Analysis of Customer Product Review

Abstract

            In recent years, sentiment analysis which is also referred as opinion mining which is the review by people about the product purchased through online, now it becomes one of the most interesting area of research. Natural Language processing plays a vital role in prediction of sentiment analysis by classifying the text into positive, negative or neutral. These increasing amounts of raw data are tremendously helpful for high source of information for automatic decision making especially in the field of sentiment analysis. Determining data polarity is the main objective of the sentiment analysis and it is the toughest challenge to optimize polarity accuracy of a sentence. This paper aims to handle the process of sentiment analysis by performing mutual information-based feature selection and adapting deep belief network and the functionality of the standard DBN is optimized by inferring the knowledge of firefly optimization to assign optimized weights on hidden nodes. The dataset used to perform this sentiment prediction of customers product review is done on Amazon dataset. Simulation results proved that this proposed model provides promising outcomes compared to the other state of arts.