BFCAI at SemEval-2022 Task 6: Multi-Layer Perceptron for Sarcasm Detection in Arabic Texts

Proceedings of the 16th International Workshop on Semantic Evaluation (SemEval-2022), pages 881 - 884 July 14-15, 2022 ©2022 Association for Computational Linguistics BFCAI at SemEval-2022 Task 6: Multi-Layer Perceptron for Sarcasm Detection in Arabic Texts Nsrin Ashraf and Fathy Elkazaz and Mohamed Taha and Hamada Nayel Department of Computer Science, Benha University, Benha, Egypt {nisrien.ashraf19,fathy.elkazzaz}@fci.bu.edu.eg {mohamed.taha, hamada.ali}@fci.bu.edu.eg Tarek Elshishtawy Department of Information System, Benha University, Benha, Egypt t.shishtawy@fci.bu.edu.eg Abstract This paper describes the systems submitted to iSarcasm shared task. The aim of iSarcasm is to identify the sarcastic contents in Arabic and English text. Our team participated in iSar- casm for the Arabic language. A multi-Layer machine learning based model has been sub- mitted for Arabic sarcasm detection. In this model, a vector space TF-IDF has been used as for feature representation. The submitted sys- tem is simple and does not need any external resources. The test results show encouraging results. 1 Introduction Analyzing social media becomes a crucial task, due to the frequently usage of social media platforms. Sarcasm detection, the conflict of using the verbal meaning of a sentence and its intended meaning (CLIFT, 1999; Lee and Katz, 1998), is an impor- tant task. Sarcasm detection is a challenge, since sarcastic contents are used to express the opposing of what is being said. Recently sarcasm detection has been studied from a computational perspec- tive as one of classification problems that separates sarcastic from non-sarcastic contents(Reyes et al., 2013; Nayel et al., 2021). Arabic is an important natural language having an extensive number of speakers. The research in Natural Language Processing (NLP) for Arabic is continually increasing. However, there is still a need to handle the complexity of NLP tasks in Ara- bic. This complexity arises from various aspects, such as orthography, morphology, dialects, short vowels, and word order. Sarcasm detection in Ara- bic is a particularly challenging task (Alayba et al., 2018). In this paper, we describe the system submitted to the iSarcasm detection shared task(Abu Farha et al., 2022). The shared task aims at detecting the sarcasm contents in Arabic tweets. In this work, a machine learning framework has been developed and various machine learning algorithms have been implemented. Term Frequency-Inverse Document Frequency (TF-IDF) has been used as vector space model for tweet representation. The rest of this paper is organized as follows: in section 2, a back- ground about sarcasm detection is given. Section 3 and section 4 overview the dataset and the sys- tem respectively. Experimental setup and results are given in section 5 and section 6 respectively. Finally, section 7 concludes the proposed work and suggests future work to be continued. 2 Background The research work have been done on Arabic sar- casm detection were mainly focused on creating datasets and establish a baseline for each created dataset (Ghanem et al., 2020). Karoui et al. (2017) created a corpus of sarcastic Arabic tweets that are related to politics. Distant supervision has been used for the creation of corpus. The authors used keywords that are …

Content truncated to 500 words