Deep Learning for Sentiment Analysis on Google Play Consumer Review

Abstract

In recent years, there has been an increasing interest in sentiment analysis on consumer reviews to understand the opinion polarity on social media. However, little attention has been paid to the development of deep learning for sentiment analysis on consumer reviews in Chinese. The research objective of this paper is to explore the impact of deep learning for sentiment analysis on Google Play consumer reviews in Chinese. A web mining technique was implemented for collecting 196,651 reviews on Google Play. We used Long Short Term Memory (LSTM) deep learning model, Naïve Bayes (NB), and support vector machine (SVM) approaches for sentiment analysis on consumer reviews and compared the experimental results. The experimental results suggest that the accuracy of deep learning for sentiment analysis on Google Play consumer review achieves 94% and deep learning approach outperforms Naïve Bayes (74.12%) and Support Vector Machine (76.46%) in the present study. Our finding confirmed that sentiment analysis on Google Play consumer review with deep learning is outstanding. The contributions of this paper are three-fold. First, the present study confirmed sentiment analysis with deep learning on Google Play consumer review may improve the accuracy of prediction. Second, we create a sentiment dictionary named iSGoPaSD for Google Play review. Third, the study compared the result of average sampling data and non-average sampling data. We found that deep learning method with non-average sampling data reached the better performance.