Building Arabic Corpora: Concepts, Methodologies, Tools, and Experiments

Both Classical and Modern Standard Arabic are understudied in Computational Linguistics and Natural Language Processing, relative to its worldwide reach as the language of the Quran, its proud heritage, and lexical richness. However, for Arabic as for many other languages, most researchers have often built corpora and develop tools that may only suit their personal objectives and for a specific time without considering the design criteria and the language features to create well-defined corpora and appropriate tools that meet the standards. In this thesis, my concern was initially to build the most needed types of Arabic corpora considering the Arabic language features but also in compliance with corpus linguistics standards. For this reason, the decision was made to investigate different methods used in previous works, develop and propose new tools and resources, adapt relevant tools to deal with the Arabic language texts, and conduct several experiences and comparative studies to provide a comprehensive description of the main stages followed while building corpora. Read more