Microblog sentiment analysis for celebrity endorsed products

Abstract

Celebrity endorsed products such as NBA basketball star player endorsed shoes are typical examples of star marketing. For the purpose of star marketing, companies need to monitor customer's emotion and sentiment to celebrities through social media such as Twitter, Instagram and Facebook. One step forward, in this study, we aim to predict sales of celebrity endorsed products using sentiment analysis on social media data on celebrities. Major fan group of celebrities are usually young generation and they use social media popularly and frequently to share their emotion on celebrities. To apply sentiment analysis on the context we propose a sentiment lexicon modification method based on supervised learning approach. Based on manually evaluated social media contents on celebrities we identify domain-specific terms and their polarities which can contribute to improve sentiment analysis performance. Using tweets on 10 NBA basketball star players and their endorsed shoes sales data we perform experiments to show the usefulness of the proposed approach.