Multi-Label Emotion Classification for Arabic Tweets

Emotion Analysis (EA)is a process of determining if the text has any emotion. EA spread significantly in the recent years, especially for social media applications as applied to tweets and Facebook posts. An assumption has been presented recently that each social media post has no intensity or has one emotion. Different cases for public posts have been considered in this work, it focuses on several emotions (multi-label)included in a single post. Tweeter posts (Tweets)have been employed to validate the proposed work, it is possible to have different intensities related to each tweet (multi-target). The proposed work focused on Arabic language tweets unlike previously implemented work, which focused on other languages such as English or Chinese. A multi-label multi-target data set of Arabic tweets annotated for emotion analysis has been built, and different experts participated in the annotation process and Cohens Kappa measure was employed to determine their concordance.