• Àüü
  • ÀüÀÚ/Àü±â
  • Åë½Å
  • ÄÄÇ»ÅÍ
´Ý±â

»çÀÌÆ®¸Ê

Loading..

Please wait....

±¹³» ³í¹®Áö

Ȩ Ȩ > ¿¬±¸¹®Çå > ±¹³» ³í¹®Áö > Çѱ¹Á¤º¸°úÇÐȸ ³í¹®Áö > Á¤º¸°úÇÐȸ ³í¹®Áö B : ¼ÒÇÁÆ®¿þ¾î ¹× ÀÀ¿ë

Á¤º¸°úÇÐȸ ³í¹®Áö B : ¼ÒÇÁÆ®¿þ¾î ¹× ÀÀ¿ë

Current Result Document : 1 / 3   ´ÙÀ½°Ç ´ÙÀ½°Ç

ÇѱÛÁ¦¸ñ(Korean Title) ÀáÀç ±¸Á¶Àû SVMÀ» È®ÀåÇÑ °áÇÕ ÇнÀ ¸ðµ¨
¿µ¹®Á¦¸ñ(English Title) Jointly Learning Model using modified Latent Structural SVM
ÀúÀÚ(Author) ÀÌâ±â   ÃÖ¼ºÇÊ   Á¤ÇѹΠ  Changki Lee   Sungpil Choi   Hanmin Jung  
¿ø¹®¼ö·Ïó(Citation) VOL 41 NO. 06 PP. 0433 ~ 0439 (2014. 06)
Çѱ۳»¿ë
(Korean Abstract)
ÀÚ¿¬¾î󸮿¡¼­´Â ¶ç¾î¾²±â ¸ðµâ°ú Ç°»ç Å°Š°°Àº ¸ðµâµéÀÌ ÆÄÀÌÇÁ¶óÀÎ ¹æ½ÄÀ¸·Î ¿¬°áµÇ¾î ¸¹ÀÌ »ç¿ëµÇ³ª, ÀÌ °æ¿ì ¾Õ ´Ü°èÀÇ ¿À·ù°¡ µÞ ´Ü°è¿¡ ´©ÀûµÇ´Â ¹®Á¦°¡ ÀÖ´Ù. º» ³í¹®¿¡¼­´Â ÆÄÀÌÇÁ¶óÀÎ ¹æ½ÄÀÇ ¹®Á¦¸¦ ÇØ°áÇϱâ À§ÇØ »ç¿ëµÇ´Â ÀϹÝÀûÀÎ °áÇÕ ¸ðµ¨À» È®ÀåÇÏ¿©, µÎ ÀÛ¾÷ÀÌ µ¿½Ã¿¡ űëµÈ ÇнÀ µ¥ÀÌÅͻӸ¸ ¾Æ´Ï¶ó ÇÑ ÀÛ¾÷¸¸ űëµÈ ÇнÀµ¥ÀÌÅ͵µ µ¿½Ã¿¡ ÇнÀ¿¡ »ç¿ëÇÒ ¼ö ÀÖ´Â °áÇÕ ÇнÀ ¸ðµ¨À» Latent Structural SVMÀ» È®ÀåÇÏ¿© Á¦¾ÈÇÑ´Ù. ½ÇÇè °á°ú, Çѱ¹¾î ¶ç¾î¾²±â¿Í ÆÄÀÌÇÁ¶óÀÎ ¹æ½ÄÀ¸·Î ¿¬°áµÈ Ç°»ç ÅÂ±ë ¼º´ÉÀº 96.77%¿´°í, ±âÁ¸ÀÇ Çѱ¹¾î ¶ç¾î¾²±â¿Í Ç°»ç ÅÂ±ë °áÇÕ ¸ðµ¨ÀÇ Ç°»ç ÅÂ±ë ¼º´ÉÀÌ 96.99%¿´À¸³ª, º» ³í¹®¿¡¼­ Á¦¾ÈÇÏ´Â °áÇÕ ÇнÀ ¸ðµ¨À» ÀÌ¿ëÇÏ¿© ´ë¿ë·®ÀÇ Çѱ¹¾î ¶ç¾î¾²±â ÇнÀµ¥ÀÌÅ͸¦ Ãß°¡·Î ÇнÀÇÑ °á°ú Ç°»ç ÅÂ±ë ¼º´ÉÀÌ 97.20%±îÁö Çâ»ó µÇ¾ú´Ù.
¿µ¹®³»¿ë
(English Abstract)
Typically, a Korean Part-of-speech (POS) tagger takes the inputs that are produced by a separate Korean word spacer. However this pipeline approach has an obvious flaw of error propagation, since the POS tagger cannot correct word spacing errors. In this paper, we describe a jointly learning model for Korean word spacing and POS tagging using modified latent structural SVM to avoid error propagation and improve word spacing by utilizing POS information and additional word spacing training data. In the case of a pipeline approach, we could achieve a 96.77% morpheme-based F-measure for POS tagging. Using the previous joint model, we could achieve a 96.99% morphemebased F-measure for POS tagging. Using the jointly learning model, we could achieve a 97.20% morpheme-based F-measure for POS tagging. Experimental results show that the jointly learning model outperforms the pipeline approach and the previous joint model.
Å°¿öµå(Keyword) °áÇÕ ÇнÀ ¸ðµ¨   Latent Structural SVM   ÆÄÀÌÇÁ¶óÀΠ  Ç°»ç ű렠 jointly learning model   latent structural SVM   pipeline   POS tagging  
ÆÄÀÏ÷ºÎ PDF ´Ù¿î·Îµå