Affiliations: Faculty of Sciences and Techniques, IR2M Laboratory, Hassan First University of Settat, Settat, Morocco
Correspondence:
[*]
Corresponding author. Soukaina Mihi, Faculty of Sciences and Techniques, IR2M Laboratory, Hassan First University of Settat, Settat, Morocco. E-mail: [email protected].
Abstract: Sentiment analysis has become a prevalent issue in the research community, with researchers employing data mining and artificial intelligence approaches to extract insights from textual data. Sentiment analysis has progressed from simply classifying evaluations as positive or negative to a sophisticated task requiring a fine-grained multimodal analysis of emotions, manifestations of sarcasm, aggression, hatred, and racism. Sarcasm occurs when the intended message differs from the literal meaning of the words employed. Generally, the content of the utterance is the opposite of the context. Sentiment analysis tasks are hampered when a sarcastic tone is recognized in user-generated content. Thus, automatic sarcasm detection in textual data dramatically impacts the performance of sentiment analysis models. This study aims to explain the basic architecture of a sarcasm detection system and the most effective techniques for extracting sarcasm. Then, for the Arabic language, determining the gap and challenges.
Keywords: Sarcasm, NLP, sentiment analysis, Arabic, deep learning, machine
learning
