An Integration of Deep Learning Approach for prediction of Consumer Reviews using Sentiment Analysis

Abstract

There is a massive increase in consumer reviews posted on social media, both in terms of the volume of evaluations and their relevance. This is why sentiment analysis is becoming increasingly popular in a wide range of contexts, including politics, social issues, and current events. For a number of corporate applications, the analysis of social media user reviews is critical. Because social media makes it so simple to obtain public sentiment from around the world, this method is better suited for sentiment mining. With the hybrid technique of two deep constructions, the Convolutional Neural Network (CNN) and the Long Short Term Memory (LSTTM), this article examines airline industry service and movie reviews (LSTM). The local feature selection power of CNN has been well documented, whereas the sequential text analysis prowess of LSTM has been well documented as well. In sentiment analysis, the suggested CNN-LSTM model primarily aims to achieve two goals. First, it can easily be scaled up to handle large amounts of social media data, and second, unlike traditional machine learning algorithm’s, it is not domain-specific. Two review datasets from different fields have been used to train a perfect that can handle all kinds of dependences in a post. This ensemble model outperforms other deep learning algorithms in accuracy and other limits, as shown by the results in the experiments.