Enhancing the performance measure of sentiment analysis through deep learning approach

Abstract

Sentiment analysis is the computational process through which one can categorize the opinions expressed in a form of text, specifically analysis is done to predict the user’s perception towards any product, movie, or any particular topic. The prediction can be negative, positive or neutral, even this can categorize to the number of class labels, it depends on the classification task. Already there are many classifiers based on machine learning and they have shown a significant result in this area, but recent years have shown a trend in deep learning due to their high performance. So this paper has also applied deep learning algorithms during the sentiment analysis. The proposed work will show how deep learning-based methods will improve the classification accuracy over machine learning algorithms. The proposed methods are evaluated on the basis of certain datasets like movie reviews, hotel reviews, and political reviews of India. In the initial phase machine learning classifiers like Naïve Bayes, Support Vector Machine, Random Forest, and logistic regression were applied later on deep learning algorithms has been implemented. With the primary focus on deep learning methods, the main contributions of numerous researchers are emphasized. The performance of the algorithm will be evaluated by the parameters like accuracy, precision, recall, and F1 score. The novelty of this work is that other than the combination of RNN and LSTM, we have implemented Deep Chrome Convolutional Neural Network for sentiment analysis and it has achieved the accuracy of 92.76% and 90.76% for a movie review and hotel review dataset respectively that is much higher than other states of the art algorithms. The analysis of sentiments will be more accurate by implementing this proposed work.