Named Entity Recognition of Persons’ Names in Arabic Tweets

Proceedings of Recent Advances in Natural Language Processing, pages 731–738, Hissar, Bulgaria, Sep 7–9 2015. Named Entity Recognition of Persons' Names in Arabic Tweets Omnia H. Zayed Center of Informatics Science Nile University Giza, Egypt omnia.zayed@gmail.com Samhaa R. El-Beltagy Center of Informatics Science Nile University Giza, Egypt samhaa@computer.org Abstract The rise in Arabic usage within various social media platforms, and notably in Twitter, has led to a growing interest in building Arabic Natural Language Processing (NLP) applica- tions capable of dealing with informal collo- quial Arabic, as it is the most commonly used form of Arabic in social media. The unique characteristics of the Arabic language make the extraction of Arabic named entities a challenging task, to which, the nature of tweets adds new dimensions. The majority of previous research done on Arabic NER fo- cused on extracting entities from the formal language, namely Modern Standard Arabic (MSA). However, the unstructured nature of the colloquial language used in tweets de- grades the performance of NER systems de- veloped to support formal MSA text. In this paper, we focus on the task of Arabic per- sons‟ names recognition. Specifically, we in- troduce an approach to extract Arabic per- sons‟ names from tweets without employing any morphological analysis or language- dependent features. The proposed approach adopts a rule-based model combined with a statistical one. This approach uses unsuper- vised learning of patterns and clustered dic- tionaries as constrains to identify a person‟s name and resolve its ambiguity. Our ap- proach outperforms the best reported result in the literature on the same test set by an in- crease of 19.6% in the F-score. 1 Introduction Named Entity Recognition (NER) is the task of identifying certain types of named expressions in unstructured text and classifying them into a pre- defined set of categories. These expressions can be personal and geographic named expressions, as well as temporal and numeric ones. NER is a crucial constituent of many Natural Language Processing (NLP) applications (Jurafsky and Martin, 2009). Examples of these applications include Machine Translation, Text Summariza- tion, Opinion Mining, and Semantic Web Searching (Benajiba et al., 2009). The advent of Twitter has offered people a significant new way of communication that ena- bles them to share their ideas, thoughts, and real- time news, an example of which was the D.C. earthquake1, which was reported on Twitter as it was unfolding. In addition, Twitter can be used by government services to reach large audiences in real time in order to send awareness messages to citizens. The sheer amount of regularly gener- ated tweets and their ubiquitous nature are among the factors that have encouraged re- searchers in many fields to analyse such content automatically for event detection and opinion mining. The informal nature of messages ex- changed within this platform poses new chal- lenges for NLP applications, as their content tends to be short, noisy and to deviate from known grammatical rules (Zayed and El-Beltagy, 2015). When it comes to automatic text analysis, the Arabic language is …

Content truncated to 500 words