ARB-SEN at SemEval-2018 Task1: A New Set of Features for Enhancing the Sentiment Intensity Prediction in Arabic Tweets

Proceedings of the 12th International Workshop on Semantic Evaluation (SemEval-2018), pages 364–368 New Orleans, Louisiana, June 5–6, 2018. ©2018 Association for Computational Linguistics ARB-SEN at SemEval-2018 Task1: A New Set of Features for Enhancing the Sentiment Intensity Prediction in Arabic Tweets El Moatez Billah Nagoudi Laboratoire d’Informatique et de Math´ematiques LIM, Laghouat, Algeria Echahid Hamma Lakhdar University, El Oued, Algeria e.nagoudi@lagh-univ.dz Abstract This article describes our proposed Arabic Sentiment Analysis system named ARB- SEN. This system is designed for the In- ternational Workshop on Semantic Eval- uation 2018 (SemEval-2018), Task1: Af- fect in Tweets. ARB-SEN proposes two su- pervised models to estimate the sentiment intensity in Arabic tweets. Both mod- els use a set of features including senti- ment lexicon, negation, word embedding and emotion symbols features. Our sys- tem combines these features to assist the sentiment analysis task. ARB-SEN sys- tem achieves a correlation score of 0.720, ranking 6th among all participants in the valence intensity regression (V-reg) for the Arabic sub-task organized within the Se- mEval 2018 evaluation campaign. 1 Introduction and Related Work According to Mohammad (2016) the Sentiment Analysis (SA) task is used to refer to the “task of automatically determining the valence or polarity of a piece of text, whether it is positive, negative, or neutral”. Nowadays, social media platforms like Twit- ter, Facebook, LinkedIn, and Quora are widely used (Lenze, 2017). For instance, Ranginwala and Towbin (2017) estimate that Twitter has 320 million active monthly users. These social me- dia platforms allow people to communicate not only the sentiment they are feeling (positive or negative) but also the intensity of this sentiment. For example, from the tweet of your friend, you can estimate that: he is very happy (most posi- tive), slightly angry (slightly negative), absolutely sad (most negative) or neutral (Mohammad et al., 2017). Automatically determining the sentiment inten- sity is an important task in several application fields, such as public health, intelligence gather- ing, commerce and social welfare (Mohammad and Bravo-Marquez, 2017). In the Semantic Evaluation (SemEval), senti- ment analysis in Twitter task has been proposed for the first time in SemEval-2013 by (Nakov et al., 2013). Since SemEval-2013, this task has be- come a principal task in SemEval : SemEval-2014 (Rosenthal et al., 2014), SemEval-2015 (Rosenthal et al., 2015), SemEval-2016, (Nakov et al., 2016) and SemEval-2017 (Rosenthal et al., 2017). In the Arab world, Salem (2017) estimates that the number of monthly active Arabic users was 11.1 million in March 2017, which makes Ara- bic an emergent language for sentiment analysis in Twitter (Rosenthal et al., 2017). Sentiment analysis task in Arabic is particularly a challeng- ing research task (Rosenthal et al., 2017) due to its complex morphological and syntactic structure (Habash, 2010). Many Arabic sentiment analysis tools and stud- ies have been proposed in order to overcome this challenge. For example in Arabic newswire, the most relevant works are: (Abdul-Mageed et al., 2011), (Elarnaoty et al., 2012). In Arabic reviews we find (Elhawary and Elfeky, 2010), (Elnagar, 2016), (Altowayan and …

Content truncated to 500 words