Sarcasm detection in microblogs using Naïve Bayes and fuzzy clustering

Abstract

Sarcasm detection of online text is a task of growing importance in the globalized world. Large corporations are interested in knowing how consumers perceive the various products launched by the companies based on analysis of microblogs, such as - Twitter, about their products.These reviews/comments/ posts are under the constant threat of being classified in the wrong category due to use of sarcasm in sentences. Automatic detection of sarcasm in microblogs, such as - Twitter, is a difficult task. It requires a system that can use some knowledge to interpret the linguistic styles of authors. In this work, we try to provide this knowledge to the system by considering different sets of features which are relatively independent of the text, namely - function words and part of speech n-grams. We test a range of different feature sets using the Naïve Bayes and fuzzy clustering algorithms. Our results show that the sarcasm detection task benefits from the inclusion of features which capture authorial style of the microblog authors. We achieve an accuracy of approximately 65% which is on the higher side of the sarcasm detection literature.