The present invention relates to statistical machine translation using parallel corpus, and more particularly, to a method and apparatus for extracting and extracting a noun phrase candidate for each language by using part-of-speech information that can constitute a noun phrase in a corpus in which a source language sentence and a target language sentence are word- A pair of noun phrases with a high probability of alignment is extracted as pairs of noun phrases in consideration of the sort probability in the pair of the noun phrases, thereby automatically extracting noun phrases from the pair of languages having poor word matching performance, And methods.
展开▼