1. UKM Learning and Research Repository
2. Master Thesis / Tesis Master
3. Faculty of Information Science and Technology / Fakulti Teknologi dan Sains Maklumat

Title:	Machine translation of classical Arabic into English using rule-based approach
Authors:	Huda Alhusain Ali Hebresha (P52689)
Supervisor:	Mohd. Juzaiddin Ab Aziz, Dr.
Keywords:	Arabic language -- Machine translating Computational linguistics Machine translating Information storage and retrieval systems Universiti Kebangsaan Malaysia -- Dissertations Dissertations, Academic -- Malaysia
Issue Date:	4-Oct-2012
Description:	Recently, Arabic machine translation has been taking place in machine translation projects and so, many researches have been carried out to improve and enhance the quality of translation from and into Arabic. This research concentrates on the translation of Classical Arabic (CA) rather than Modern Standard Arabic. The problems of this research are the difficulty of delivering the appropriate meaning of the Classical Arabic term in the target language (English), different sentence structure between two languages, and word agreement and ordering problem. The main aim of our work is to create an Arabic-English machine translation system to translate text from Classical Arabic (CA) into English using rule-based approach. To achieve this objective; first, we modified a lexical database to contain vocabulary which related to the environment of Classical Arabic. Then we build and implemented a machine translation prototype called (CAMT) to overcome the difficulties of translation of Classical Arabic to English. Our system includes three phases which are analysis, transfer and generation phase. Firstly, in the analysis phase the CA input text is analysed morphologically and syntactically. Morphological analysis produces information about the part-of-speech of each word in the CA input text and its sub-category such as (Number, gender, tense, etc.). The work of the syntactical analysis is to check the correctness of sentence grammatically and in spelling. Secondly, in transfer stage we use the information from previous phase to acquire reasonable rule of the CA input text structure and its equivalent rule in the target language (English). The translation of each word must be chosen from the dictionary in this step. In the final phase, using the information from the transfer stage the Arabic source text will be generated to obtain the target text in English. The evaluation process of this work is done by comparing the output produced by our system with the original human translation. We have carried out two experiments to evaluate CAMT system which are: automatic evaluation using iBLEU metric and human judgment evaluation. The 1-gram results produced by iBLEU algorithm show that the accuracy of CAMT is nearly 89.4%, while human judgment evaluation results gave CAMT system around 91.0% accuracy. The results produced by both experiments prove that the system provides good results.,Certification of Master's / Doctoral Thesis" is not available
Pages:	132
Call Number:	Z699.H384 2012 3 tesis
Publisher:	UKM, Bangi
Appears in Collections:	Faculty of Information Science and Technology / Fakulti Teknologi dan Sains Maklumat

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.