Abstract
This chapter presented an analysis of the application of lexicon-based political sentiment analysis in social media. The aim is to identify the most frequently used lexicons in political sentiment analysis, their results, similarities, and differences. For this, the authors conducted a systematic literature review based on PRISMA methodology. Afinn, NRC, and SenticNet lexicons are tested and combined for data analysis from the 2020 U.S. presidential campaign. Findings show that political sentiment analysis is a new field studied for only 10 years. Political sentiment analysis could generate benefits in understanding problems such as political polarization, discourse analysis, politician influence, candidate profiling, and improving government-citizen interaction, among other problems in the public sphere, enhanced by the combination of lexicons and multimodal analysis. The authors conclude that polarity was one of the critical dimensions identified for finding variations in the behavior and polarity of sentiments. Limitations and future work also are presented.
TopIntroduction
Opinion mining and sentiment analysis has been applied to different areas of knowledge such as marketing, business, finance, and political contexts (Bing et al., 2012; Charalabidis et al., 2015; Valle-Cruz, Lopez-Chau, et al., 2021; Vinodhini & Chandrasekaran, 2012; Wang et al., 2020). Sentiment analysis allows identifying polarity in social media posts, inherent emotions in texts, images, videos, facial, and body expressions. Specifically, in social networks such as Twitter and Facebook, the analysis of polarity has been widely studied. However, some other classifications and models allow a multimodal analysis to determine different dimensions of emotional charges (Cambria et al., 2020; Valle-Cruz, Lopez-Chau, et al., 2021). Sentiment analysis, a neuralgic technique in natural language processing (Bing, 2012), is useful for social media analysis in different kinds of applications. These applications include prediction, profiling, emotion, sentiment analysis, polarity, and preference detection (Chau et al., 2021; López-Chau et al., 2020; Sandoval-Almazan & Valle-Cruz, 2020; Valle-Cruz, Fernandez-Cortez, et al., 2021), among others. Regarding the political context studies, from different perspectives, such as public policy, public administration, political campaigns, political communication, and discourse ideology; the analysis of conviction, polarization, as well as the emotions and biases generated are vital to understanding the phenomenon under study and its possible consequences (Rhodes, 2014; Riggs, 1965). In recent decades, the debate generated in social media around the political sphere has gained importance. Throughout the world, candidates use social networks to interact with citizens, express their ideology and promote their campaign promises. Although it is still necessary to campaign in the traditional way - not only virtually - but it has also become relevant to monitor the users' behavior towards the events that happen in the political scene. Political sentiment analysis makes it possible to identify the moods, sentiments, emotions, preferences, and impressions of potential voters in an electoral campaign (Anwar et al., 2021; Sandoval-Almazan & Valle-Cruz, 2020; Valle-Cruz, Lopez-Chau, et al., 2021).
Key Terms in this Chapter
Polarity: Value assigned to a word or phrase, resulting from lexicon-based sentiment analysis. The assigned values can be negative and positive, or the absolute value of negativity or positivity.
Social Networks: Tools used to disseminate information and interact with people virtually. Some examples are Twitter, Facebook, Instagram, and Tik Tok.
MFUL: The most frequently used lexicons for sentiment analysis.
Lexicon: Vocabulary of a person, language, or branch of knowledge. Lexicon-based sentiment analysis is based on a set of words labeled as positive, negative, and neutral sentiments.
Emotion: Intense feeling of joy, sadness, anger (among others) produced by an event, an idea, or a memory. Lexicon-based sentiment analysis allows approaching the measurement of emotions.
Social media: Online communication platforms where content is created by the users themselves using Web 2.0, such as blogs, social networks, instant messaging, and wikis.
Political Sentiment Analysis: Detection of polarity and emotions in political content. The content can be text, images, or videos.