Pengembangan Example Based Machine Translation dengan Sistem Tasrif pada Sebuah Computer Aided Translation Arab-Indonesia

Hudaya, Kharisman Kholid (2009) Pengembangan Example Based Machine Translation dengan Sistem Tasrif pada Sebuah Computer Aided Translation Arab-Indonesia. Masters thesis, Institut Teknologi Sepuluh Nopember Surabaya.

Text 5105201018-Master_Thesis.pdf - Accepted Version Download (9MB)

Abstract

Example Based Machine Translation (EBMT) adalah sebuah paradigma mesin penerjemahan yang menggunakan sumber corpus contoh terjemahan. Paradigma ini lebih baru dibandingkan dua paradigma lain: Rule Based Machine Translation dan Statistical Machine Translation. Tingkat akurasi output terjemahan dari EBMT tergantung dari kelengkapan cakupan corpus contoh terjemahan. Kemampuan EBMT dibatasi dengan jumlah dan jangkauan dari training datanya. Bahkan dengan corpus yang sangat besar, kita tidak akan memiliki contoh terjemahan yang mencakup segala kemungkinan kalimat yang ingin diterjemahkan. Kualitas basil terjemahan suatu kalimat yang dihasilkan mesin penerjemah seringkali jauh dari harapan. Karena itu, penerjemahan dengan bantuan computer (Computer Aided Translation) lebih dipilih, karena manusia memiliki lebih banyak kontrol dalam mengklasifikasikan suatu kata berdasarkan makna kontekstualnya. Penelitian ini bertujuan mengembangkan suatu prototipe Computer Aided Translation Arab - Indonesia yang mengadaptasi paradigma EBMT, ditunjang dengan Sistem Tashrif (Tashrif System), sebuah metode baru yang diperkenalkan. Uji coba terhadap literatur berbahasa Arab dalam domain Ilmu Fiqh menunjukkan presisi 71% sedangkan pada portal berita online AlJazeera menunjukkan presisi 42%
====================================================================================================================================
Example Based Machine Translation (EBMT) is a machine translation paradigm that uses a corpus of translation examples. This paradigm is newer than the other two paradigms: Rule Based Machine Translation and Statistical Machine Translation. The level of accuracy of the translation output from EBMT depends on the completeness of the translation sample corpus coverage. EBMT's capabilities are limited by the amount and range of training data. Even with a very large corpus, we will not have translation examples that cover all possible sentences we want to translate. The quality of the translation results for a sentence produced by a translation machine is often far from expectations. Therefore, computer aided translation (Computer Aided Translation) is preferred, because humans have more control in classifying a word based on its contextual meaning. This research aims to develop a prototype of Arabic - Indonesian Computer Aided Translation that adapts the EBMT paradigm, supported by the Tashrif System, a new method that has been introduced. Tests on Arabic language literature in the Fiqh Science domain showed 71% precision, while the online news portal AlJazeera showed 42% precision

Item Type:	Thesis (Masters)
Additional Information:	RTIf 005.45 Hud p-1 2008
Uncontrolled Keywords:	EBMT, Computer Aided Translation, Arab-Indonesia, Machine Translation
Subjects:	Q Science > QA Mathematics > QA76.6 Computer programming. Q Science > QA Mathematics > QA76.9.U83 Graphical user interfaces. User interfaces (Computer systems)--Design.
Divisions:	Faculty of Information Technology > Informatics Engineering > 55101-(S2) Master Thesis
Depositing User:	EKO BUDI RAHARJO
Date Deposited:	01 Jul 2024 04:13
Last Modified:	01 Jul 2024 04:13
URI:	http://repository.its.ac.id/id/eprint/108096

Actions (login required)

View Item