Twitter Sentiment Classification Using Binary Shuffled Frog Algorithm

Abstract

AbstractTwitter is a popular social networking site allowing users to read/post messages (tweets). Among the topic varieties, people in Twitter express sentiments for brands, stars, merchandises, and social events. Hence, it draws attention to assess a crowd’s sentiments in Twitter. Tweets classify a target’s sentiments as positive, negative or neutral. Individuals comment on many entities (or targets) in a tweet, thereby affecting availabilities for current methods. This is beneficial for clients who explore products sentiment before acquisition, or corporations wanting to check public sentiment of their products. This work proposes a new Twitter Sentiment Classification algorithm using novel feature selection technique with ensemble classifier through a meta-heuristic algorithm. Feature vectors are represented using binary encoding and a novel transfer function to flip encoding bits using shuffled frog meta-heuristic algorithm is proposed. To evaluate the new algorithm, Twitter corpus from Stanford Univ...