Building an Arabic Transliterator and Annotating Data Morphologically : Using Weka to Annotate our Arabic Corpus Morphologically and Compare it with the Xerox Arabic Analyser's Results.

ISBN-13: 9783330847750 / Angielski / Miękka / 2017 / 56 str.

Abdulaziz Al Jumaia

Opis

In this book, we, firstly, discuss related works to ours. Secondly, we create a transliteration program, produce our own corpus, use the Xerox Arabic analyser to morphologically annotate a raw Arabic text, use Weka to train our transliterated corpus, and then, compare the annotation of the Xerox analyser with the results of Weka. The book shows the methods used to create our own transliteration system using a dictionary which maps the Arabic letters with the Latin letters. To do that, we use a raw Arabic text taken from a chapter of the book "Al-Bidayah Wan-Nihayah" for Ibn Kathir and store the results for a later use. the book progresses to discuss the use of the same original text, used previously for transliteration, in the Xerox Arabic analyser which uses a finite-state transducer to annotate the text morphologically. The annotations are, then, selected manually (gold-standard), added to our transliterated text and trained using different algorithms in Weka. Ultimately, the results of Weka are compared with the gold-standard annotation.

Widoczna okładka, to zdjęcie poglądowe, a rzeczywista szata graficzna może różnić się od prezentowanej. Nie gwarantujemy zgodności okładki z prezentowanym zdjęciem.

O autorze

Abdulaziz bin Yaqoob Yousef Al Jumaia was born in Al Ahsa, Saudi Arabia in 1991. He has a bachelor degree in English linguistics and a master's degree in language and information processing. He is interested in Machine Learning, Machine Translation, Corpus Linguistics and Natural Language Processing.