NULL at SemEval-2022 Task 6: Intended Sarcasm Detection Using Stylistically Fused Contextualized Representation and Deep Learning

Proceedings of the 16th International Workshop on Semantic Evaluation (SemEval-2022), pages 862 - 870 July 14-15, 2022 ©2022 Association for Computational Linguistics NULL at SemEval-2022 Task 6: Intended Sarcasm Detection Using Stylistically Fused Contextualized Representation and Deep Learning Mostafa Rahgouy Department of Computer Science Auburn University, Alabama, USA mzr0108@auburn.edu Hamed Babaei Giglou Department of Computer Science University of Tabriz, Tabriz, Iran h.babaei98@ms.tabrizu.ac.ir Taher Rahgooy Department of Computer Science University of West Florida, Florida, USA trahgooy@students.uwf.edu Cheryl D Seals Department of Computer Science Auburn University, Alabama, USA sealscd@auburn.edu Abstract The intended sarcasm cannot be understood un- til the listener observes that the text’s literal meaning violates truthfulness. Consequently, words and meanings play an essential role in specifying sarcasm. Enriched feature extrac- tion techniques were proposed to capture both words and meanings in the contexts. Due to the overlapping features in sarcastic and non- sarcastic texts, a CNN model extracts local fea- tures from the combined class-dependent sta- tistical embedding of sarcastic texts with con- textualized embedding. Another component BiLSTM extracts long dependencies from com- bined non-sarcastic statistical and contextual- ized embeddings. This work combines a classi- fier that uses the combined high-level features of CNN and BiLSTM for sarcasm detection to produce the final predictions. The experimen- tal analysis presented in this paper shows the effectiveness of the proposed method. 1 Introduction Sarcasm detection is a specific case of sentiment analysis where the focus is on detecting sarcasm in text. Therefore, the task is to detect if a given text is sarcastic or not. According to (Hacker, 2011), sarcasm refers to the use of words that mean the opposite of what you want to say, especially to in- sult someone, show irritation, or provide humor. However, (Oprea and Magdy, 2020; Wilson, 2006) define sarcasm as a form of irony marked by a dis- crepancy between the literal and intended meanings of an utterance, through which the speaker usually manifests a hostile, derogatory, or contemptuous attitude. Generally, sarcasm tweets/texts are the utterances of a positive statement with harmful in- tent, and since the intent is hard to detect not only for computers but also for humans, it has attracted a considerable body of research in the natural lan- guage processing field to study opinions given by the users of online social media platforms such as Twitter, Facebook, Reddit, and Instagram. In the SemEval 2022 at Task 6: iSarcasmEval (Abu Farha et al., 2022), the goal is to identify intended sar- casm in both English and Arabic languages. This task consists of 3 subtasks, and we will focus on SubTask A defined as: SubTask A: Given a text, determine whether it is sarcastic or non-sarcastic. Since sarcasm is an utterance of a statement with negative intent (e.g., He has the best taste in music.”) and an admiring tone (e.g., ”She al- ways makes dry jokes.”). We hypothesized that contextualized representation might not be enough to represent intent when considering the sentiment analysis challenge. Therefore, we combined the representation of stylistically statistical and con- …

Content truncated to 500 words