High Tech team at SemEval-2022 Task 6: Intended Sarcasm Detection for Arabic texts

Proceedings of the 16th International Workshop on Semantic Evaluation (SemEval-2022), pages 840 - 843 July 14-15, 2022 ©2022 Association for Computational Linguistics High Tech team at SemEval-2022 Task 6: Intended Sarcasm Detection for Arabic texts Alami Hamza1, Abdessamad Benlahbib2, Ahmed Alami3, 1 Innov-Tech Laboratory, Departement of Engineering, High Technology School, Rabat, Morocco 2 LISAC Laboratory, Faculty of Sciences Dhar EL Mehraz (F.S.D.M), Sidi Mohamed Ben Abdellah University (U.S.M.B.A) 3 Ibn Tofail University, National School of Applied Sciences, Kenitra, Morocco hamza0alami@gmail.com, abdessamad.benlahbib@usmba.ac.ma, alami.alami1996@gmail.com Abstract This paper presents our proposed methods for iSarcasmEval shared task. The shared task consists of three different subtasks. We par- ticipate in both subtask A and subtask C. The purpose of the subtask A was to predict if a text is sarcastic while the aim of subtask C is to determine which text is sarcastic given a sarcastic text and its non-sarcastic rephrase. Both of the developed solutions used BERT pre-trained models. The proposed models are optimized on simple objectives and easy to grasp. However, despite their simplicity our methods ranked 4 and 2 in iSarcasmEval sub- task A and subtask C for Arabic texts. 1 Introduction Nowadays, social media users provide a huge amount of text, images and videos. This large amount of data contains useful information (users ideas, opinions, events, etc) for various domains such as stock predictions, marketing, or politics. In order to benefit from these data, new fields of study have been introduced including sentiment analysis, opinion mining, author profiling, and harassment detection (Liu, 2012; Rosenthal et al., 2014; May- nard and Greenwood, 2014; Van Hee et al., 2018). Natural Language Processing (NLP) algorithms are used extensively in these fields to extract useful in- formation. For instance, to determine whether a given product has a positive or negative sentiment in the market, we can apply NLP techniques to analyse a list of twitter posts to infer a sentiment about the product. According to Oxford dictionary, sarcasm is “The use of irony to mock or convey centempt”. Sarcastic text convey negative implied sentiment, however it can have positive, negative, or no surface senti- ment. Sarcasm is commonly used in social media, thus it introduces errors in various tasks such as sentiment analysis and opinion mining. This is explained in the work of Rosenthal et al. (2014), it shows a significant drop in sentiment polarity classification performance when processing sarcas- tic tweets, compared to non-sarcastic ones. In this context, the task iSarcasmEval: Intended Sarcasm Detection In English and Arabic (Abu Farha et al., 2022) is organized by SemEval 2022. The main tasks consists of three subtask: • Subtask A: Given a text, determine whether it is sarcastic or non-sarcastic. • SubTask B (English only): A binary multi- label classification task. Given a text, deter- mine which ironic speech category it belongs to, if any. • SubTask C: Given a sarcastic text and its non- sarcastic rephrase, i.e. two texts that convey the same meaning, determine which is the sarcastic one. In this paper, we describe …

Content truncated to 500 words