Exploring the Realization of Irony in Twitter Data

Exploring the Realization of Irony in Twitter Data Cynthia Van Hee, Els Lefever and V´eronique Hoste LT3, Language and Translation Technology Team Ghent University Groot-Brittanni¨elaan 45, 9000 Ghent, Belgium cynthia.vanhee, els.lefever, veronique.hoste@ugent.be Abstract Handling figurative language like irony is currently a challenging task in natural language processing. Since irony is commonly used in user-generated content, its presence can significantly undermine accurate analysis of opinions and sentiment in such texts. Understanding irony is therefore important if we want to push the state-of-the-art in tasks such as sentiment analysis. In this research, we present the construction of a Twitter dataset for two languages, being English and Dutch, and the development of new guidelines for the annotation of verbal irony in social media texts. Furthermore, we present some statistics on the annotated corpora, from which we can conclude that the detection of contrasting evaluations might be a good indicator for recognizing irony. Keywords: social media, figurative language processing, verbal irony 1. Introduction With the arrival of Web 2.0, technologies like social me- dia have become accessible to a vast amount of people. As a result, they have become valuable sources of information about the public’s opinion. What characterizes social media content is that it is often rich in figurative language (May- nard and Greenwood, 2014; Reyes et al., 2013). Han- dling figurative language represents, however, one of the most challenging tasks in natural language processing. It is often characterized by linguistic devices such as humor, metaphor and irony, whose meaning goes beyond the literal meaning and is therefore often hard to capture, even for hu- mans. Effectively, understanding figurative language often requires world knowledge and familiarity with the conver- sational context and the cultural background of the conver- sation’s participants; information that is difficult to access by machines. Verbal irony is a particular genre of figurative language that is traditionally defined as saying the opposite of what is actually meant (Curc´o, 2007; Grice, 1975; Mc- Quarrie and Mick, 1996). Evidently, this type of figurative language has important implications for tasks that handle subjective information, such as sentiment analysis (May- nard and Greenwood, 2014). Sentiment analysis involves the automatic extraction of positive and negative opinions from online texts. It goes without saying that its accuracy can be significantly undermined by the presence of irony, as illustrated by Example 1. (extracted from the SemEval- 2015 training corpus for Task 111). (1) It was so nice of my dad to come to my grad- uation party. #not Regular sentiment analysis systems will probably classify this tweet as positive, whereas the intended emotion is un- deniably a negative one. The hashtag #not indicates the presence of irony in this example. By contrast, in exam- ple 2. (taken from Riloff et al. (2013)), there is no explicit indication of irony present. Nevertheless, the irony is no- ticeable because given our world knowledge, we know that the act of going to the dentist (for a root canal) is typically 1http://alt.qcri.org/semeval2015/task11 an unpleasant situation. This clearly contrasts with the pos- …

Content truncated to 500 words