Survey of Sentiment Analysis Using Deep Learning Techniques

Abstract

This paper presents a detailed review of deep learning techniques used in Sentiment Analysis. Sentiment analysis is one of the most researched areas in natural language processing. Natural language processing has a wide range of applications like voice recognition, machine translation, product review, aspect oriented product analysis, sentiment analysis and text classification like email categorization and spam filtering. The conventional methods used for sentiment analysis is lexicon based processing. However, with the advancements in the field of artificial intelligence, the machine learning algorithms started to play a major role in sentiment analysis applications. Currently deep learning technique is the latest hotspot being used for predicting the sentiments. Several research works have been carried out in the Natural Language Processing (NLP) using the deep learning methods. The most popular deep learning methods employed includes Convolution Neural Network (CNN) and Recurrent Neural Network (RNN) particularly the Long Short Term Memory (LSTM). These techniques are used in combination or as stand-alone based on the domain area of application. The focus of this survey is on the various flavors of the deep learning methods used in different applications of sentiment analysis at sentence level and aspect/target level. Furthermore, the advantages and drawbacks of the methods are discussed along with their performance parameters.