Á¤º¸°úÇÐȸ ³í¹®Áö B : ¼ÒÇÁÆ®¿þ¾î ¹× ÀÀ¿ë
Current Result Document :
ÇѱÛÁ¦¸ñ(Korean Title) |
ÀÏ-ÇÑ ±â°è ¹ø¿ª ½Ã½ºÅÛ : ¿¬¾î ÆÐÅÏÀ» ÀÌ¿ëÇÑ ¾îÈÖ ´ÙÀǼº ÇØ¼Ò |
¿µ¹®Á¦¸ñ(English Title) |
Resolution of Word Sense Ambiguities using Collocation Patterns in Japanese-to-Korean MT System |
ÀúÀÚ(Author) |
¹®°æÈñ
ÀÌÁ¾Çõ
±èÁ¤ÀÎ
¾ç±âÁÖ
Kyong-Hi Moon
Jong-Hyeok Lee
Jung-In Kim
Gijoo Yang
|
¿ø¹®¼ö·Ïó(Citation) |
VOL 25 NO. 08 PP. 1270 ~ 1280 (1998. 08) |
Çѱ۳»¿ë (Korean Abstract) |
º» ³í¹®¿¡¼´Â ¹®¹ý ü°èÀÇ À¯»ç¼ºÀ¸·Î ÀÎÇØ ÁÖ·Î Á÷Á¢ ¹ø¿ª ¹æ½ÄÀ» äÅÃÇÏ°í ÀÖ´Â ÀÏ-ÇÑ ±â°è ¹ø¿ª¿¡¼ °íÇ°ÁúÀÇ ¹ø¿ªÀ» À§ÇÑ ¾îÈÖ ´ÙÀǼº ÇØ¼Ò ±â¹ýÀ» Á¦¾ÈÇÑ´Ù. ±¸¹®, ÀÇ¹Ì Á¤º¸°¡ »ó´ëÀûÀ¸·Î ºÎÁ·ÇÑ Á÷Á¢ ¹ø¿ª ¹æ½Ä¿¡¼´Â ³ôÀº ¼öÁØÀÇ ¾îÈÖ ´ÙÀǼº ÇؼҸ¦ À§ÇØ Á¤±³ÇÑ °ø±â Á¤º¸°¡ ¿ä±¸µÈ´Ù. µû¶ó¼ ´ÙÀǼºÀ» °¡Áö´Â ¾îÈֵ鿡 ´ëÇØ ´Ù¾çÇÑ ÇüÅÂÀÇ °ø±â Á¦¾àÀ» ÀÚÀ¯·Ó°Ô Ç¥ÇöÇÒ ¼ö ÀÖ´Â ¿¬¾î ÆÐÅÏÀÇ ±â¼ú¹æ¹ýÀ» Á¦¾ÈÇÏ°í, ¸í»ç¿Í ¿ë¾ðÀÇ ½Ã¼Ò·¯½º °èÃþ ±¸Á¶¸¦ ÀÌ¿ëÇÑ À¯»çµµ Á¡¼ö °è»ê¹ýÀ» ÅëÇØ °¡Àå ÀûÀýÇÑ Àǹ̰¡ ¼±ÅõǴ º¯È¯ ¸ðµ¨À» Á¦½ÃÇÑ´Ù. ¼º´É Æò°¡¸¦ À§ÇÏ¿© ´ÙÀǾîÀÇ ´ëºÎºÐÀ» Â÷ÁöÇÏ´Â Á¶»ç¿Í µ¿»ç¸¦ ´ë»óÀ¸·Î ½ÇÇèÇÑ °á°ú, Á¶»ç¿¡ ´ëÇؼ´Â ¾à 93.3%, µ¿»ç¿¡ ´ëÇؼ´Â ¾à 82.6%ÀÇ ³ôÀº ´ÙÀǼº ÇؼÒÀ²À» º¸ÀÓÀ¸·Î½á Á¦¾ÈÇÑ ¹æ¹ýÀÌ À¯È¿ÇÔÀ» º¸¿´´Ù.
|
¿µ¹®³»¿ë (English Abstract) |
Due to grammatical similarities of both languages, almost all Japanese-Korean MT systems have been developed under the direct MT strategy, in which the word sense disambiguation is considered essential to high-quality translation, but also as a difficult task due to lack of syntactic and semantic clues. In this paper, a word sense disambiguation using sophisticated collocation patterns is proposed under the direct MT framework. First, we present a description method of collocation patterns that can specify a variety of selection restrictions between words. And then we also describe a similarity scoring mechanism based on thesaurus hierarchies of nouns and verbs to select the best-matched one from among several possible patterns. Finally, some experimental evaluations were done using a large-scale corpus of Asahi newspaper articles. They show that the word sense ambiguity of highly polysemous particles and verbs can be resolved with an accuracy of 93.3% and 82.6% for each. |
Å°¿öµå(Keyword) |
|
ÆÄÀÏ÷ºÎ |
PDF ´Ù¿î·Îµå
|