Á¤º¸Ã³¸®ÇÐȸ ³í¹®Áö ¼ÒÇÁÆ®¿þ¾î ¹× µ¥ÀÌÅÍ °øÇÐ
Current Result Document :
ÇѱÛÁ¦¸ñ(Korean Title) |
½ºÅÃ-Æ÷ÀÎÅÍ ³×Æ®¿öÅ©¿Í ºÎºÐ Æ®¸® Á¤º¸¸¦ ÀÌ¿ëÇÑ Çѱ¹¾î ÀÇÁ¸ ±¸¹® ºÐ¼® |
¿µ¹®Á¦¸ñ(English Title) |
Korean Dependency Parsing Using Stack-Pointer Networks and Subtree Information |
ÀúÀÚ(Author) |
°Çö¾Æ
ÀÓÈñ¼®
Hyun-Ah Kang
Heui-Seok Lim
ÃÖ¿ë¼®
ÀÌ°øÁÖ
Yong-Seok Choi
Kong Joo Lee
|
¿ø¹®¼ö·Ïó(Citation) |
VOL 10 NO. 06 PP. 0235 ~ 0242 (2021. 06) |
Çѱ۳»¿ë (Korean Abstract) |
º» ¿¬±¸¿¡¼´Â Æ÷ÀÎÅÍ ³×Æ®¿öÅ© ¸ðµ¨À» ÀÇÁ¸ ±¸¹® ºÐ¼®¿¡ ¸Â°Ô È®ÀåÇÑ ½ºÅÃ-Æ÷ÀÎÅÍ ³×Æ®¿öÅ© ¸ðµ¨À» ÀÌ¿ëÇÏ¿© Çѱ¹¾î ÀÇÁ¸ ±¸¹® ºÐ¼®±â¸¦ ±¸ÇöÇÑ´Ù. ½ºÅÃ-Æ÷ÀÎÅÍ ³×Æ®¿öÅ© ¸ðµ¨ ±â¹Ý ÀÇÁ¸ ±¸¹® ºÐ¼®±â´Â ÀÎÄÚ´õ-µðÄÚ´õ·Î ±¸¼ºµÇ¾î ÀÖÀ¸¸ç ´Ù¸¥ ÀÇÁ¸ ±¸¹® ºÐ¼®±â¿Í ´Þ¸® ³»ºÎ ½ºÅÃÀ» °®°í ÀÖ¾î ·çÆ®ºÎÅÍ ½ÃÀÛÇÏ´Â ÇÏÇâ½Ä ±¸¹® ºÐ¼®ÀÌ °¡´ÉÇÏ´Ù. µðÄÚ´õÀÇ °¢ ´Ü°è¿¡¼´Â ÀÇÁ¸¼Ò¸¦ ã±â À§ÇØ ºÎ¸ð ³ëµå»Ó¸¸ ¾Æ´Ï¶ó ÀÌ¹Ì ÆÄ»ýµÈ Æ®¸® ±¸Á¶¿¡¼ Á¶ºÎ¸ð¿Í ÇüÁ¦ ³ëµå¸¦ ÂüÁ¶ÇÒ ¼ö ÀÖ´Ù. ±âÁ¸ ¿¬±¸¿¡¼´Â ´Ü¼øÇÏ°Ô ÇØ´ç ³ëµåµéÀÇ ÇÕÀ» °è»êÇÏ¿© ÀÔ·ÂÀ¸·Î »ç¿ëÇÏ¿´°í, ÇüÁ¦ ³ëµåÀÇ °æ¿ì¿¡´Â °¡Àå ÃÖ±Ù¿¡ ¹æ¹®Çß´ø °Í¸¸À» »ç¿ëÇÒ ¼ö ÀÖ¾ú´Ù. º» ¿¬±¸¿¡¼´Â ±×·¡ÇÁ ¾îÅÙ¼Ç ³×Æ®¿öÅ©¸¦ µµÀÔÇÏ¿© ÀÌ¹Ì ÆÄ»ýµÈ ºÎºÐ Æ®¸®¸¦ Ç¥ÇöÇÏ°í À̸¦ ½ºÅÃ-Æ÷ÀÎÅÍ ³×Æ®¿öÅ©ÀÇ ÀÔ·ÂÀ¸·Î »ç¿ëÇϵµ·Ï ±¸¹® ºÐ¼®±â¸¦ ¼öÁ¤ÇÑ´Ù. ¼¼Á¾ ÄÚÆÛ½º¿Í ¸ðµÎÀÇ ÄÚÆÛ½º¸¦ ´ë»óÀ» ½ÇÇèÇÑ °á°ú ·¹À̾î 2ÀÇ ±×·¡ÇÁ ¾îÅÙ¼Ç ³×Æ®¿öÅ©¸¦ ÀÌ¿ëÇÏ¿© ºÎºÐ Æ®¸®¸¦ Ç¥ÇöÇßÀ» ¶§ ƯÈ÷ ¹®Àå ´ÜÀ§ÀÇ ±¸¹® ºÐ¼® Á¤È®µµ¿¡¼ ¸¹Àº ¼º´É Çâ»óÀ» È®ÀÎÇÒ ¼ö ÀÖ¾ú´Ù. |
¿µ¹®³»¿ë (English Abstract) |
In this work, we develop a Korean dependency parser based on a stack-pointer network that consists of a pointer network and an internal stack. The parser has an encoder and decoder and builds a dependency tree for an input sentence in a depth-first manner. The encoder of the parser encodes an input sentence, and the decoder selects a child for the word at the top of the stack at each step. Since the parser has the internal stack where a search path is stored, the parser can utilize information of previously derived subtrees when selecting a child node. Previous studies used only a grandparent and the most recently visited sibling without considering a subtree structure. In this paper, we introduce graph attention networks that can represent a previously derived subtree. Then we modify our parser based on the stack-pointer network to utilize subtree information produced by the graph attention networks. After training the dependency parser using Sejong and Everyone¡¯s corpus, we evaluate the parser¡¯s performance. Experimental results show that the proposed parser achieves better performance than the previous approaches at sentence-level accuracies when adopting 2-depth graph attention networks. |
Å°¿öµå(Keyword) |
°Ë»ö ÁúÀÇ À¯Çü
ÅؽºÆ®¸¶ÀÌ´×
ÅäÇȸ𵨸µ
PCA
·Î±× ºÐ¼®
Search Query Types
Text Mining
Topic Modeling
PCA
Log Analysis
½ºÅÃ-Æ÷ÀÎÅÍ ³×Æ®¿öÅ©
Çѱ¹¾î ÀÇÁ¸ ±¸¹® ºÐ¼®
±×·¡ÇÁ ¾îÅÙ¼Ç ³×Æ®¿öÅ©
ºÎºÐ Æ®¸®
»çÀü ÈÆ·ÃµÈ ´Ü¾î Ç¥Çö
¸ðµÎÀÇ ÄÚÆÛ½º
Stack-Pointer Networks
Korean Dependency Parser
Graph Attention Networks
Subtree
Pre-trained Word Representation Model
Everyone¡¯s Corpus
|
ÆÄÀÏ÷ºÎ |
PDF ´Ù¿î·Îµå
|