2003 Fiscal Year Final Research Report Summary
Machine Translation Using Semantically Classified Expression Structure Pattern Pairs
| Project/Area Number |
13480091
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (B)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
Intelligent informatics
|
|---|
| Research Institution | NIIGATA UNIVERSITY |
|---|
Principal Investigator |
MIYAZAKI Masahiro NIIGATA UNIVERSITY, faculty of Engineering, Professor, 工学部, 教授 (80221612)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
NITTA Yoshihiko Nihon University, College of Economics, Professor, 経済学部, 教授 (20277370)
SHIGATA Katuyuki Fukuoka University, Faculty of Science, Professor, 工学部, 教授 (50028213)
IKEHARA Satoru Tottori University, faculty of Engineering, Professor, 工学部, 教授 (70283968)
IKEDA Takashi Gifu University, faculty of Engineering, Professor, 工学部, 教授 (10232183)
|
|---|
| Project Period (FY) |
2001 – 2003
|
| Keywords | Machine Translation / Pattern Translation / Semantic Typology / Analogical Mapping |
|---|
| Research Abstract |
Recognizing the thoughts of semantic typology that "the way of object recognition is reflected in the structure of linguistic expression", we research machine translation method that absorbs the differences in recognition between languages through the semantic classification of expressions of source language and target language.
The results of research are as follows.
1.Research and building of semantically classified Japanese-English structure pattern pairs
(1)Using the technique of N-gram statistics, we developed methods to automatically extract Japanese-English expression structure patterns of complex, compound, and special structure sentences from large scale Japanese-English translation samples.
(2)We designed the script language for describing Japanese-English expression structure patters and we checked its description of patterns.
(3)We designed the meta-languages for describing the semantic typology and we checked its semantically typological expression capability.
(4)Based on the technique of ATN, we designed the method to match between the given Japanese/English expressions and the large scale Japanese/English structure patterns with high speed.
(5)We built Japanese-English expression structure pattern pairs of complex, compound, and special structure sentences.
2.Research and prototyping Japanese-English Machine translation system
(1)We have researched the mechanism of Japanese-English machine translation methods that translate by applying the existing Japanese-English expression structure pattern pairs of simple the existing Japanese-English expression structure pattern pairs of simple sentences and semantically classified Japanese-English expression structure pattern pairs of complex, compound, and special structure sentences.
(2)We verified the possibility and effectiveness as the machine translation methods by prototyping Japanese-English Machine translation system.
|
