Role of Emotion icons in Sentiment classification of Arabic Tweets

TLDR: An automatic approach to predict sentiments for informal Arabic language is proposed and it is observed that although emotion icons presence in the tweets helps in development of comparatively more accurate classifier, however they play ambiguous role in defining the sentiments of tweets.

Abstract: In recent years, there is enormous increase of data content due to emergence of social media platforms in digital word of internet. The text mining is very important technique to discover the knowledge from unstructured data. Automatic sentiment analysis is one of the important applications of text mining. The sentiment analysis is used to predict the text polarity (positive, negative, and neutral). Furthermore, the most of users using social media such as Twitter use informal language to express their opinions. In this paper, we propose an automatic approach to predict sentiments for informal Arabic language. We chose Arabic tweets as input for our study. We observed through our experiment results that although emotion icons presence in the tweets helps in development of comparatively more accurate classifier, however they play ambiguous role in defining the sentiments of tweets.