불-한 병렬 말뭉치의 문장 단위 정렬 방식에 대한 연구
A Study on the Sentence-Level Alignment Method of the French-Korean Parallel Corpus
2
조회
0
다운로드
4
인용
0
참고문헌
키워드
초록
A French-Korean parallel corpus consists of a source text written in French and a target text that is a Korean translation of the source text. It can be used for machine translation, automatic information extraction on the Internet, comparative linguistics, etc. The aim of this paper is to introduce French-Korean bilingual parallel corpora and, in particular, to describe the linguistic problems in the alignment process, particularly on the sentence level. Bilingual parallel corpora will first be characterized and compared with monolingual reference corpora, after which the linguistic problems that are usually met during the alignment process, particularly on the sentence level, will be described, such as various correspondence types, ambiguous sentence segmentation, and structural differences between languages. Finally, the statistical methods that are applied in the construction of parallel corpora and the linguistic hypothesis these are based on will be examined, and some arguments will be presented to show that these methods are not suitable for French-Korean parallel corpora.
Abstract
A French-Korean parallel corpus consists of a source text written in French and a target text that is a Korean translation of the source text. It can be used for machine translation, automatic information extraction on the Internet, comparative linguistics, etc. The aim of this paper is to introduce French-Korean bilingual parallel corpora and, in particular, to describe the linguistic problems in the alignment process, particularly on the sentence level. Bilingual parallel corpora will first be characterized and compared with monolingual reference corpora, after which the linguistic problems that are usually met during the alignment process, particularly on the sentence level, will be described, such as various correspondence types, ambiguous sentence segmentation, and structural differences between languages. Finally, the statistical methods that are applied in the construction of parallel corpora and the linguistic hypothesis these are based on will be examined, and some arguments will be presented to show that these methods are not suitable for French-Korean parallel corpora.