Popularity Prediction in Twitter During Financial Events

Abstract

Twitter is one of the widely used micro blogging services around the world. During major events people tend to share posts, comments and links creating a tremendous amount of tweets. The retweet feature provided by Twitter can be used as a filtering mechanism and to measure the popularity of a tweet. Popularity prediction in Twitter has been widely approached in the literature. However, not much has been done in the area of finance. This work is a preliminary step toward understanding the characteristics of finance related tweets. A small scale experiment is carried out to investigate the tweet’s features that could influence finance related tweets popularity. Using these features, a prediction model was created using binary logistic regression. The research concludes that not all features are created equal when it comes to popularity in the financial context. The research found that some features highly influence popularity such as the verified feature, while other features such follower count of the tweeter does not directly influence tweet’s popularity.