On tweets, retweets, hashtags and user profiles in the 2016 American Presidential Election Scene

Abstract

Twitter is a microblogging where users can publish short messages restricted to 140 characters. It has been used in the political scene from different perspectives. One of them is predicting election results. In this area, many researchers have drawn their attention to hashtag studies. However, its use is still limited to the collection and selection stages, related to the prediction process. In addition, most studies investigating hashtags have performed an arbitrary hashtag selection. Tweets/retweets are still the main source of information to prediction election results. In this paper, the relevance of hashtags available on tweets / retweets and on the descriptions of user's profiles was investigated. Furthermore, descriptions of user's profiles were investigated to verify whether the political position expressed by users is relevant in a presidential sample. In order to do so, 1,974,401 tweets / retweets from 432,289 different users were collected during the 2016 presidential election campaign in the US. The main conclusion revealed that the most frequent hashtags contained first names, surnames and candidates' campaign slogans; 10% of all messages had a political hashtag, and users expressing some kind of political position in their descriptions posted 20.7% of all messages.