Sentiment analysis for product rating using a deep learning approach
Krishna Kumar Mohbey
Abstract
Sentiment analysis is a study about opinions, emotions, and attitudes of the people towards an event or issue. Social networking is an invaluable medium for individuals to express their thoughts and views about any subject or topic, contributing to massive quantities of unstructured knowledge. These emotions can be processed and examined to analyze and obtain insights. Therefore, several machine learning and Natural Language Processing (NLP) based methods were used to examine these opinions. Because of the shifts in the sequential order, stream lengths, and complex logic, the exact sentiments in the consumer feedback are still difficult. Recently, deep learning methods have been used to attain improved results. The most common forms of deep learning method used are the Recurrent Neural Network (RNN) and the Convolutional Neural Network (CNN). Long short-term memory (LSTM), particularly with attentive layers, pays greater focus to the sentiment impact. LSTM has an advantage over alternative RNNs and other deep learning approaches because of relative insensitivity to gap length. Deep learning approaches have better accuracy over existing state-of-the-art approaches because of their ability to handle extensive realtime data and the power of feature extraction, but there is still room for improvement. This paper has used Long Short-Term Model (LSTM) model to predict the customer review's opinion, attaining an accuracy of 93.66%. Furthermore, comparative analysis of the deep LSTM model with existing has been presented.