Combining Humor and Sarcasm for Improving Political Parody Detection

Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 1800 - 1807 July 10-15, 2022 ©2022 Association for Computational Linguistics Combining Humor and Sarcasm for Improving Political Parody Detection Xiao Aoα Danae Sánchez Villegasα Daniel Preo¸tiuc-Pietroβ Nikolaos Aletrasα α Computer Science Department, University of Sheffield, UK β Bloomberg {xao3,dsanchezvillegas1,n.aletras}@sheffield.ac.uk dpreotiucpie@bloomberg.net Abstract Parody is a figurative device used for mimick- ing entities for comedic or critical purposes. Parody is intentionally humorous and often involves sarcasm. This paper explores jointly modelling these figurative tropes with the goal of improving performance of political parody detection in tweets. To this end, we present a multi-encoder model that combines three paral- lel encoders to enrich parody-specific represen- tations with humor and sarcasm information. Experiments on a publicly available data set of political parody tweets demonstrate that our approach outperforms previous state-of-the-art methods.1 1 Introduction Parody is a figurative device which imitates enti- ties such as politicians and celebrities by copying their particular style or a situation where the en- tity was involved (Rose, 1993). It is an intrinsic part of social media as a relatively new comedic form (Vis, 2013). A very popular type of parody is political parody, which is used to express politi- cal opposition and civic engagement (Davis et al., 2018). One of the hallmarks of parody expression is the deployment of other figurative devices, such as hu- mor and sarcasm, as emphasized on studies of par- ody in linguistics (Haiman et al., 1998; Highfield, 2016). For example, in Table 1 the text expresses sarcasm about Myspace2 being a ‘winning tech- nology’, while mocking the fact that three more popular social media sites were unavailable. This example also highlights the similarities between parody and real tweets, which may pose issues to misinformation classification systems (Mu and Ale- tras, 2020). 1Code is available here https://github. com/iamoscar1/Multi_Encoder_Model_for_ Political_Parody_Prediction 2https://myspace.com Twitter Handle @Queen_UK Parody tweet Boris Johnson on the phone. Very smug that #myspace hasn’t gone down. Says he’s always backed winning technologies #whatsappdown #instagramdown #FacebookIsDown Table 1: Example of a parody tweet3 by the Twitter han- dle @Queen_UK. Humor and sarcasm are expressed simultaneously. These figurative devices have so far been stud- ied in isolation to parody. Previous work on mod- eling humor in computational linguistics has fo- cused on identifying jokes, i.e., short comedic passages that end with a hilarious line (Hetzron, 1991), based on linguistic features (Taylor and Ma- zlack, 2004; Purandare and Litman, 2006; Kid- don and Brun, 2011) and deep learning techniques (Chen and Soo, 2018; Weller and Seppi, 2019; An- namoradnejad and Zoghi, 2020). Similarly, compu- tational approaches for modeling sarcasm (i.e., a form of verbal irony used to mock or convey con- tent) in texts have been explored (Davidov et al., 2010; González-Ibáñez et al., 2011; Liebrecht et al., 2013; Rajadesingan et al., 2015; Ghosh et al., 2020, 2021), including multi-modal utterances, i.e. texts, images, and videos (Cai et al., 2019; Castro et al., 2019; Oprea and …

Content truncated to 500 words