Semantic Role Labeling of Arabic Emotional Text in Tweets

Abstract

We propose an Arabic corpus annotated with semantic role labels and emotion to improve Arabic NLP tasks.Emotion analysis and semantic role labeling concern many areas of applications and represent a big challenge for NLP tasks. To the best of our knowledge,few studies in the literature have attempted to integrate semantic role labeling with emotion analysis. However, Arabic language suffers from a lack of such a relevant datasets and tools compared to English. In this research, we build a corpus of 3000 Arabic tweets annotated with emotion categories and their related arguments using a semi-automatic tool that supports Frame semantics based annotation. This ongoing effort represents a first stepto providinga sizeable annotated corpus for the Arabic Language.