Work

Foreign Words and the Automatic Processing of Arabic Social Media Text Written in Roman Script

Year: 2014

Type: conference-paper

Abstract: Arabic on social media has all the properties of any language on social media that make it tough for natural language processing, plus some specific problems.These include diglossia, the use of an alternative alphabet (Roman), and code switching with foreign languages.In this paper, we present a system which can process Arabic written in Roman alphabet ("Arabizi").It identifies whether each word is a foreign word or one of another four categories (Arabic, name, punctuation, sound), and transliterates Arabic words and names into the Arabic alphabet.We obtain an overall system performance of 83.8% on an unseen test set.

Source: Workshop on Computational Approaches to Code Switching

DOI: 10.3115/v1/w14-3901

Authors: Ramy Eskander, Mohamed Al-Badrashiny, Nizar Habash, Owen Rambow

Institutions: Columbia University, George Washington University, New York University Abu Dhabi

Language: English

Topic: Natural Language Processing Techniques

Subfield: Artificial Intelligence

Field: Computer Science

Domain: Physical Sciences

Open Access status: hybrid

Funder: Defense Advanced Research Projects Agency

OpenAlex ID: W2250967669