Identifying the Overlap between Election Result and Candidates’ Ranking Based on Hashtag-Enhanced, Lexicon-Based Sentiment Analysis

The popularity and availability of Twitter as a service and a data source have fueled the interest in sentiment analysis. Previous research has shed light on the challenges that contextualizing effects and linguistic complexities pose for the accurate sentiment classification of tweets. We test the effect of adding manually-annotated, corpus-based hashtags to a sentiment lexicon, finding that this step in combination with negation detection increases prediction accuracy by about 7%. We then use our enhanced model to identify and rank the candidates of the Republican and Democratic Party of the 2016 New York primary election by the decreasing ratio of tweets that mentioned these individuals and had positive valence, and compare our results to the election outcome.