A Study and Comparison of Sentiment Analysis Techniques Using Demonetization: Case Study

Introduction

The aim of sentiment analysis (SA) is to discover the emotional orientation of user opinion or review of the particular topic, product or object. Sentiment analysis is different from traditional text mining in that we focus on a particular topic for mining whereas sentimental analysis is much complex. Sentimental analysis can be considered as a classification process in which opinion or reviews are classified as positive, negative or neutral (Lei et al., 2016). Sentimental analysis has played an important role in various applications of text mining for consumer opinion detection, customer relationship management, brand and product poisoning and reviewed analysis. Based on the sentimental analysis multiple companies can know the status of their product and take different decision to enhance their business.

There are a lot of research has been done in the field of sentimental analysis (Medhat et al., 2014) which is mostly classify reviews based on their polarity. In general, a review can be categorized by different topics, for example, mobile phone reviews most likely discuss topic of a feature such as a price, brand, memory, camera, etc. in sentimental analysis task, users are not only interested in the opinion of review but also interested in the topics discovery. Therefore, it is known that the sentiment polarities are also dependent on various topics. The complete process of sentiment classification is shown in Figure 1.

The classification of sentiments can be performed on three levels such as document, sentence and aspect level. Document-level SA classifies an opinion as positive or negative. Sentence-level SA expresses each sentence as positive or negative opinion. When classifying opinion as document or sentence level, there is no significant difference because sentences are the short documents (Liu, 2012). While aspect level SA classifies entities or objects concerning specific aspect. For example, a different user may have different aspects towards mobile phone such as voice quality, battery life, camera quality, etc. To classify sentiments, it is needed to have a review or opinion information. There are different sources from where the user can collect data for sentiment analysis. Microblogs and social networks are excellent resources for information in these days because they provide an efficient environment to share opinions.

The contribution of this chapter is significant in different dimensions. It provides different applications of SA and fields where it can play an important role. This chapter also describes various SA techniques for opinion classification which is used to make decisions and policy making. Different SA fields such as emotion detection and transfer learning also discussed in this chapter. Finally, this chapter provides various case studies of sentiment analysis. The proposed chapter includes the concept of sentiment analysis, different techniques for sentiment analysis and case studies for opinion mining. It discusses various applications of sentiment analysis and classification in different areas.

Figure 1.

Sentiment analysis process

This chapter is organized as follows: section 2 includes various applications of sentimental analysis. Section 3 discusses various sentiment classification approaches. In section 4, several feature extraction methods are included that is required for sentiment analysis. Section 5 presents different fields where sentiment analysis used. Sentiment analysis case studies are added in section 6. Lastly, discussion and conclusions are presented in section 7 and 8, respectively.

Motivation

Sentiment analysis or opinion mining is the computational study of people’s opinion, thought, attitudes and emotions toward an object. The object can represent individuals, events, topics, tweet or review. Sentiment analysis identifies the attitude of people expressed in a text then analyses it. The expression of people may belong to acceptance, rejection or neutral regarding any event.