:: Kasetsart University ::

หน้าแรก > งานวิจัย > Plaesarn: Machine Aided Translation Tool for English-to-Thai

Plaesarn: Machine Aided Translation Tool for English-to-Thai

เธฃเธจ.เธ”เธฃ. อัศนีย์ ก่อตระกูล

1. Abstract

Internet technology nowadays influences Thailand's country development. Since many relevant information and knowledge, spreading on the Internet, is mostly written in English, Thai people can therefore slowly acknowledge them. This engineering project's major aim is to develop a web-based software tool in order to assist document translation from English to Thai through using Structural Transfer Model as the mechanism. The system's architecture namely comprises of two parts: server part and client part. The server part consists of four sub-systems, Syntactic Analysis System, Structural Transformation System, Sentence Generation System, and Linguistic Knowledge Acquisition System. The client part is an Internet browser. The system primary advantage is to assist Thai people to read English documents better. Its major features are when an English sentence can be translated to many translations; the user can manually elect which of them is the most appropriate for him, and the user can manually teach the system a new translation rule if considered necessary. The principal limitation is that the system translates English sentences regardless of semantic ambiguity management. For this project, three translation problems, namely, word rearrangement, verbal phrase translation, and Thai noun classifier identification, are accomplished, but, in contrast, semantic ambiguity is still not achieved. Moreover, the system is experimented on the Future Magazine Corpus and it yields 75% of translation accuracy.

2. Objectives

The project is developed with the following three major purposes:

To study Automatic Machine Translation theories and existing Machine Translation system
To study relevant problems occurring in English-to-Thai translation
To design and implement a software tool to assist English-to-Thai translation

3. System Architecture The Machine Translation System comprises of four principle components, namely:

Syntactic Analysis: This stage analyzes every sentence of the source documents through using grammatical knowledge from the Linguistic Knowledge Base. Then, it transforms each sentence into a parse tree, which is selected as a sentence representation in order to relieve complicatedness of translation.
Structural Transformation: This stage transforms a source language parse tree into a target language one through using grammatical knowledge from the Linguistic Knowledge Base.
Sentence Generation: This stage generates a target language sentence from the parse tree. This stage also relies on the grammatical knowledge from the Linguistic Knowledge Base.
Linguistic Knowledge Acquisition: This stage provides advantageous tools for training translation knowledge.

4. Figures

Kasetsart University was established on 2 February 1943 with the prime aims in promoting subjects related to agricultural sciences.

The university revised its curricula and expanded the subject areas to cover science, arts, social science, humanity, education, engineering, and architecture.

Recently, the university made an attempt to include medicine and health science.

Kasetsart University has established 7 campuses distributed scatterly to cover all regions of Thailand.

At present, the number of enrolled students at all levels of study is 23,000.

Figure 1: The input paragraph

Figure 2: This is the homepage of the translation service

Figure 3: The user can manually choose the most appropriate translation.

Kasetsart University ถูกก่อตั้งเมื่อ 2 February 1943 ด้วยจุดประสงค์หลักในการสนับสนุนวิชาซึ่งถูกเกี่ยวข้องถึงวิทยาศาสตร์การเกษตร

มหาวิทยาลัยทบทวนหลักสูตรและขยายพื้นที่วิชาเพื่อครอบคลุมวิทยาศาสตร์, ศิลปศาสตร์, สังคมวิทยาศาสตร์, มนุษยศาสตร์, การศึกษา, วิศวกรรม, และสถาปัตยกรรม

เมื่อเร็ววันนี้, มหาวิทยาลัยทำความพยายามเพื่อรวมแพทยศาสตร์และวิทยาศาสตร์สุขภาพ

Kasetsart University ได้ก่อตั้งวิทยาเขต 7 วิทยาเขตซึ่งถูกกระจายอย่างกระจัดกระจายเพื่อครอบคลุมพื้นที่ทั้งหมดของ Thailand

ที่ปัจจุบัน, จำนวนของนักเรียนที่เข้าเรียนที่ระดับชั้นทั้งหมดของการศึกษาเป็น 23,000

Figure 4: The output paragraph

Figure 5: The user can manually train to the system new translation rules.

Kasetsart University of Thailand

หน้าแรก || ค้นหาข้อมูล || ติดต่อ
Copyright (C) 2002. Kasetsart University
Revised : June, 2002
Contact WebMasTeR of Kasetsart University