Title: The impact of political party/candidate on the election results from a sentiment analysis perspective using #AnambraDecides2017 tweets

Abstract: This work investigates empirically the impact of political party control over its candidates or vice versa on winning an election using a natural language processing technique called sentiment analysis (SA). To do this, a set of 7430 tweets bearing or related to #AnambraDecides2017 was streamed during the November 18, 2017, Anambra State gubernatorial election. These are Twitter discussions on the top five political parties and their candidates termed political actors in this paper. We conduct polarity and subjectivity sentiment analyses on all the tweets considering time as a useful dimension of SA. Furthermore, we use the word frequency to find words most associated with the political actors in a given time. We find most talked about topics using a topic modeling algorithm and how the computed sentiments and most frequent words are related to the topics per political actor. Among other things, we deduced from the experimental results that even though a political party serves as a platform that sales the personality of a candidate, the acceptance of the candidate/party adds to the winning of an election. For example, we found the winner of the election Willie Obiano benefiting from the values his party share among the people of the State. Associating his name with his party, All Progressive Grand Alliance (APGA) displays more positive sentiments and the subjective sentiment analysis indicates that Twitter users mentioning APGA are less emotionally subjective in their tweets than the other parties.