(±¸)Á¤º¸°úÇÐȸ ³í¹®Áö
Current Result Document : 1 / 1
ÇѱÛÁ¦¸ñ(Korean Title) |
Àº´Ð ¸¶¸£ÄÚÇÁ ¸ðµ¨À» ÀÌ¿ëÇÑ È¿À²ÀûÀÎ Çѱ¹¾î Ç°»çÀÇ Å±ë |
¿µ¹®Á¦¸ñ(English Title) |
An Efficient Korean Part-of-Speech Tagging using a Hidden Markov Model |
ÀúÀÚ(Author) |
±èÀçÈÆ
ÀÓö¼ö
¼Á¤¿¬
JaeHoon Kim
ChulSu Lim
JungYun Seo
|
¿ø¹®¼ö·Ïó(Citation) |
VOL 22 NO. 01 PP. 0136 ~ 0146 (1995. 01) |
Çѱ۳»¿ë (Korean Abstract) |
Ç°»ç űëÀº ÀÚ¿¬¾ð¾î ó¸® ºÐ¾ß¿¡¼ Ç°»çÀÇ ¸ðÈ£¼º ÇؼҴ ¹°·Ð À½¼®ÀÎ½Ä ¹× ¹®ÀÚÀνÄÀÇ ÈÄ󸮱â·Î¼ ¸¹ÀÌ »ç¿ëµµ°í ÀÖ´Ù. º» ³í¹®Àº Àº´Ð ¸¶¸£ÄÚÇÁ ¸ðµ¨Àº ÀÌ¿ëÇÑ Çѱ¹¾î Ç°»ç ű뿡 °üÇؼ ³íÇÑ´Ù. Çѱ¹¾î´Â ¿µ¾î¿Í´Â ´Þ¸® Ç°»ç űëÀ» ÇÒ ¶§¿¡ ¾îÀýÀÌ ºÐ¸® µÇ¾î¾ß ÇÑ´Ù. ¾îÁú ºÐ¸®µÉ ¶§¿¡ Çѱ¹¾îÀÇ °æ¿ì¿¡ ¿©·¯ Á¾·ù·Î ºÐ¸®µÇ¾î À̸¦ Àº´Ð ¸¶¸£ÄÚÇÁ ¸ðµ¨¿¡ Àû¿ëÇÒ °æ¿ì¿¡ ´ÙÀԷ¿(´ÙÀÔ·Â ´Ü¾î¿) ¹®Á¦°¡ ¹ß»ýµÈ´Ù. º» ³í¹®¿¡¼´Â Àº´Ð ¸¶¸£ÄÚÇÁ ¸ðµ¨À» ÀÌ¿ëÇÏ¿© Çѱ¹¾î Ç°»ç ÅÂ±ë ¹®Á¦¸¦ Ç® ¶§¿¡ ÀÌ¿Í °°Àº ¹®Á¦ÀÇ ÇØ°á¹æ¹ýÀ» Á¦½ÃÇÏ°í, À̵鿡 ´ëÇÑ È¿À²ÀûÀÎ ¹æ¹ýÀ» Á¦½ÃÇÑ´Ù. Çѱ¹¾î Ç°»ç űëÀ» À§ÇÑ Àº´Ð ¸¶¸£ÄÚÇÁ ¸ðµ¨ÀÇ ÇнÀÀº ¿Ã¹Ù¸¥ ÇüżҸ¦ ºÐ¸®ÇÑ ÇнÀ ¸»¹¶Ä¡¸¦ ¸¸µêÀ¸·Î½á ÇØ°áÇÏ°í, Ç°»ç¿(»óÅ¿) ã±â´Â °øÀ¯ ´Ü¾î¿ÀÇ °³³äÀ» ÀÌ¿ëÇÏ¿© ÇØ°áÇÏ°í °¡»ó ´Ü¾î °³³äÀ» ÀÌ¿ëÇÏ¿© Á» ´õ È¿À²ÀûÀÎ ¹æ¹ýÀ» Á¦½ÃÇÑ´Ù. ¶Ç Çѱ¹¾î ¹®Àå¿¡ ´ëÇÑ Ç°»ç űëÀ» ½ÇÁ¦ Àû¿ë ½ÇÇèÇÏ¿© Á¦½ÃÇÑ ¹æ¹ýÀÇ À¯¿ë¼ºÀ» º¸¿´´Ù.
|
¿µ¹®³»¿ë (English Abstract) |
Applications of a part-of-speech tagging system include ambiguity resolution in natural language processing, a post-processing in speech recognition and character recognition and so on. In this paper, we describe a Korean part-of-speech tagging system. In Korean, most word phrases are made up of more than one morpheme. However, one word phrase may be analyze in several different ways due to morphological ambiguities. Furthermore, each analyzed result may consist of different numbers of morphemes. It causes multiple observation(word) sequences in the hidden Markov model for part-of-speech tagging. To alleviate the problems, in this paper we suggest a method for assigning a part-or-speech tag to each morpheme in Korean. The method is based on a hidden Markov model which can be trained without using any tagged corpus. To relax the amount of computation to process multiple word sequences, which are extraordinarily occurred in Korean part -of-speech tagging, we develop a revised Viterbi algorithm for determining the most promising tag sequence using shared word sequences and virtual observarions(words). Experimental results show that the model in useful in Korean part-of-speech tagging.
|
Å°¿öµå(Keyword) |
|
ÆÄÀÏ÷ºÎ |
PDF ´Ù¿î·Îµå
|