Pathways for irony detection in tweets
Larissa Astrogildo de Freitas, Aline Aver Vanin, Denise Nauderer Hogetop, Marco N. Bochernitsan, Renata La Rocca Vieira · 2014
Posts on Twitter allow users to express ideas and opinions in a very dynamic way. The volume of data available is incredibly high in this support, and it may provide relevant clues regarding the judgement of the public on certain product, event, service etc. While in standard sentiment analysis the most common task is to classify the utterances according to their polarity, it is clear that detecting ironic senses represent a big challenge for Natural Language Processing. By observing a corpus constitued by tweets, we propose a set of patterns that might suggest ironic/sarcastic statements. Thus, we developed special clues for irony detection, through the implementation and evaluation of a set of patterns.