POS Tagging for Arabic Tweets

Proceedings of Recent Advances in Natural Language Processing, pages 1–8, Hissar, Bulgaria, Sep 7–9 2015. POS Tagging for Arabic Tweets Fahad Albogamy School of Computer Science, University of Manchester, Manchester, M13 9PL, UK albogamf@cs.man.ac.uk Allan Ramsay School of Computer Science, University of Manchester, Manchester, M13 9PL, UK allan.ramsay@cs.man.ac.uk Abstract Part-of-Speech (POS) tagging is a key step in many NLP algorithms. However, tweets are difficult to POS tag because there are many phenomena that frequently appear in Twitter that are not as common, or are en- tirely absent, in other domains: tweets are short, are not always written maintaining formal grammar and proper spelling, and abbreviations are often used to overcome their restricted lengths. Arabic tweets also show a further range of linguistic phenom- ena such as usage of different dialects, romanised Arabic and borrowing foreign words. In this paper, we present an evalu- ation and a detailed error analysis of state- of-the-art POS taggers for Arabic when applied to Arabic tweets. The accuracy of standard Arabic taggers is typically excel- lent (96-97%) on Modern Standard Arabic (MSA) text; however, their accuracy de- clines to 49-65% on Arabic tweets. Fur- ther, we present our initial approach to im- prove the taggers’ performance. By doing some improvements based on observed er- rors, we are able to reach 79% tagging ac- curacy. 1 Introduction The last few years have seen an enormous growth in the use of social networking platforms such as Twitter in the Arab World. A study prepared and published by Semiocast in 2012 has revealed that Arabic was the fastest growing language on Twit- ter in 2011. People post about their lives, share opinions on a variety of topics and discuss current issues. There are millions of tweets daily, yielding a corpus which is noisy and informal, but which is sometimes informative. As a result, Twitter has become one of the most important social informa- tion mutual platforms. The nature of the text con- tent of microblogs differs from traditional blogs. In Twitter, for example, a tweet is short and con- tains a maximum of 140 characters. Tweets also are not always written maintaining formal gram- mar and proper spelling. They are ambiguous and rich in acronyms. Slang and abbreviations are of- ten used to overcome their restricted lengths (Java et al., 2007). POS tagging is an essential processing step in a wide range of high level text processing appli- cations such as information extraction, machine translation and sentiment analysis (Barbosa and Feng, 2010). However, people working on Ara- bic tweets have tended to concentrate on low level lexical relations which were used for shallow pars- ing and sentiment analysis such as (Mourad and Darwish, 2013; El-Fishawy et al., 2014). They do not use the standard linguistic pipeline tools such as POS tagging which might enable a richer lin- guistic analysis (Gimpel et al., 2011). The prop- erties listed above of the microblogging domain make POS tagging on Twitter very different from their counterparts in more formal texts. It is …

Content truncated to 500 words