Sarcasm Detection in Libyan Arabic Dialects Using Natural Language Processing Techniques

Sarcasm is often used to convey the opposite of what is actually said (positive words, negative meaning). It is a type of humor that relies on the listener or reader to understand the intended meaning of the words. Sarcasm can be difficult to detect in written text, especially on social media platforms where people often use abbreviated language and emojis to convey their messages. The sarcasm detection can be a crucial role in refining Automated Sentiment (SA). In order to enhance the effectiveness of data-driven approaches for customer service and product development. Furthermore, the existing research on sarcasm detection in Arabic is limited. Also, there is a noticeable absence of studies in particular, the sarcasm detection in Libyan Arabic dialects. This study utilizes a data-set of 5082 collected from Facebook platform, labeled by team annotators experienced in the Libyan dialects as (sarcastic or non-sarcastic), and extracting some lexical and syntactic features which used to training Support Vector Machine (SVM) model. The results show commendable performance with an accuracy of 79.15%, precision of 79.3%, recall of 79.7%, and an F1 score of 79.5%.