Comprehensive Analysis of Classifier to Identify Sentiment in Football Specific Tweets

Abstract

The football fans' feelings get unfold during the different phases of the football match and they express their emotions, stance on Twitter. This research work focuses on identifying the stance expressed by the fans on Twitter. The changes in fans' opinions are reflected in a series of tweets written by fans. In this paper, various classification algorithms are used to analyze and categorize sentiments present in tweets posted during the 2018 FIFA World Cup. In this work, a football-specific dataset is created and labeled manually. From the dataset, a lexicon related to football-specific sentiment is created. We use domain-specific lexicons, the TF-IDF feature selection method, Count Vectorizer, and various sentiment classifiers to identify the sentiment expressed by football fans on Twitter. In this paper, the performance of different classifier algorithms is analyzed while determining the hidden sentiment. Our experiment results demonstrate that the Random Forest algorithm exhibits consistent and robust performance compared to other classifiers