AraSenTi: Large-Scale Twitter-Specific Arabic Sentiment Lexicons

Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics, pages 697–705, Berlin, Germany, August 7-12, 2016. c⃝2016 Association for Computational Linguistics AraSenTi: Large-Scale Twitter-Specific Arabic Sentiment Lexicons Nora Al-Twairesh1,2, Hend Al-Khalifa2, AbdulMalik Al-Salman1 Computer Science Department1, Information Technology Department2 College of Computer and Information Sciences King Saud University {twairesh,hendk,salman@ksu.edu.sa} Abstract Sentiment Analysis (SA) is an active research area nowadays due to the tremendous interest in aggregating and evaluating opinions being disseminated by users on the Web. SA of English has been thoroughly researched; however research on SA of Arabic has just flourished. Twitter is considered a powerful tool for disseminating information and a rich resource for opinionated text containing views on many different topics. In this paper we attempt to bridge a gap in Arabic SA of Twitter which is the lack of sentiment lexi- cons that are tailored for the informal lan- guage of Twitter. We generate two lexicons extracted from a large dataset of tweets using two approaches and evaluate their use in a simple lexicon based method. The evaluation is performed on internal and external da- tasets. The performance of these automatical- ly generated lexicons was very promising, al- beit the simple method used for classification. The best F-score obtained was 89.58% on the internal dataset and 63.1-64.7% on the exter- nal datasets. 1 Introduction The past decade has witnessed the proliferation of social media websites which has led to the production of vast amounts of unstructured text on the Web. This text can be characterized as objective, i.e. containing facts, or subjective i.e. containing opinions and sentiments about enti- ties. Sentiment Analysis (SA) is the research field that is concerned with identifying opinions in text and classifying them as positive, negative or neutral. SA of English has been thoroughly researched; however research on SA of Arabic has just flourished. Arabic is ranked fourth among languages on the web although it is the fastest growing lan- guage on the web among other languages (Inter- net World Stats, 2015). Arabic is a morphologi- cally rich language where one lemma can have hundreds of surface forms; this complicates the tasks of SA. Moreover, the Arabic language has many variants. The formal language is called Modern Standard Arabic (MSA) and the spoken language differs in different Arabic countries producing numerous Arabic dialects sometimes called informal Arabic or colloquial Arabic. The language used in social media is known to be highly dialectal (Darwish and Magdy, 2014). Dialects differ from MSA phonologically, mor- phologically and syntactically and they do not have standard orthographies (Habash, 2010). Consequently, resources built for MSA cannot be adapted to dialects very well. The informal language used in social media and in Twitter in particular makes the SA of tweets a challenging task. The language on social media is known to contain slang, nonstandard spellings and evolves by time. As such sentiment lexicons that are built from standard dictionaries cannot adequately capture the informal language in social media text. Therefore, in this paper we propose …

Content truncated to 500 words