EMOTION DETECTION IN ARABIC TEXT USING MACHINE LEARNING METHODS

Abstract

Emotions are essential to any or all languages and are notoriously challenging to grasp. The textual data with embedded emotions has increased considerably with the Internet and social networking platforms. This study aims to tackle the challenging problem of emotion detection in Arabic text. Recent studies found that dialect diversity and morpho-logical complexity in the Arabic language, with limited access to annotated training datasets for Arabic emotions, pose the foremost significant challenges to Arabic emotion detection. The previous few years have seen a giant increase in interest in text emotion detection. The study of Arabic emotions might be a result of the Arab world’s considerable influence on global politics and thus the economy. There are numerous uses for the automated recognition of emotions within the textual content on Facebook and Twitter, including company development, program design, content generation, and emergency response. Hence, we shall develop a machine-learning model for emotion detection from Arabic textual data on social platforms. This study categorizes the texts supported emotions, anger, joy, sadness, and fear, using supervised machine learning approaches. We used five different machine learning algorithms, namely Decision Tree (DT), K-Nearest Neighbor (KNN), Naive Bayes (NB), Multinomial Naive Bayes (NB), and Support Vector Machine (SVM) to classify emotions in Arabic tweets. These results found that the selection Tree and K- Nearest Neighbor classifiers have the simplest accomplishment regarding the accuracy, 0.74, While the NB and Multinomial NB classifiers acquired 0.69, and also the SVM obtained 0.63.