Deep Learning Architectures Based Sentiment Analysis Systematic Literature Review

Abstract

Recently, sentiment analysis has attracted a good deal of interest from researchers. Each day, FB, Twitter, Weibo, and other social media, in addition to huge e-trade websites, generate big feedback, which can be used to investigate humans' critiques or emotions. Sentiment analysis will correctly analyze subjective records from this information. Deep learning becomes an effective machine learning approach which learns from a couple of layers of representations or features of the records and outputs a cutting-edge prediction effect. Deep learning overcomes all other machine learning techniques for sentiment analysis as well as offers different architectures for sentiment evaluation. Recently, RNN abbreviated as Recurrent Neural Networks as well as CNN abbreviated as Convolutional Neural Networks has been replaced with Transformer Language models in deep learning architecture.This survey deals with contemporary architectures and challenges in that are in Sentiment Analysis as well as Natural Language Processing.