Political marketing with data analytics

Social media played a significant role in past Presidential elections, and it is very likely that this form of communication will continue to influence political campaigns. Can analytics uncover the linguistic “plot arcs” and resulting sentiment or emotion in political text? This paper examines how natural language processing (NLP) and data visualization tools and methods in analytics can play a key role in marketing political candidates. Using publicly available text messages, the authors employ NLP techniques to transform the text observations from the past campaigns of Hillary Clinton, Barack Obama, and Donald Trump into a linguistic “corpus” and story arc visualizations. The methodology includes the use of Syuzhet and Latent Dirichlet Allocation (LDA) models. The resulting data visualizations reveal the story arcs associated with the candidate’s communications, and they provide a means to analyze the unbiased political sentiment or hidden emotion in the text. In an analysis of the results, the authors found distinctly different story arcs and vocabulary usage among the three Presidential candidates. The contribution to the literature is a methodology for extracting the story and the resulting sentiment from text messages for marketing campaigns. The authors suggest that the techniques used in this paper can assist future research on marketing other products or services that utilize computer-mediated communications.