Application of Deep Learning Approaches for Sentiment Analysis

Abstract

Social media platforms, forums, blogs, and opinion sites generate vast amount of data. Such data in the form of opinions, emotions, and views about services, politics, and products are characterized by unstructured format. End users, business industries, and politicians are highly influenced by sentiments of the people expressed on social media platforms. Therefore, extracting, analyzing, summarizing, and predicting the sentiments from large unstructured data needs automated sentiment analysis. Sentiment analysis is an automated process of extracting the opinionated from data and classifying the sentiments as positive, negative, and neutral. Lack of enough labeled data for sentiment analysis is one of the crucial challenges in natural language processing. Deep learning has emanated as one of the highly sought-after solutions to address this challenge due to automated and hierarchical learning capability inherently supported by deep learning models. Considering the application of deep learning approaches for sentiment analysis, this chapter aims to put forth taxonomy of traits to be considered for deep learning-based sentiment analysis and demystify the role of deep learning approaches for sentiment analysis.