Á¤º¸°úÇÐȸ ³í¹®Áö B : ¼ÒÇÁÆ®¿þ¾î ¹× ÀÀ¿ë
ÇѱÛÁ¦¸ñ(Korean Title) |
¿ÂÅç·ÎÁö ½ºÅ°¸¶ Æ®¸®ÇÃÀÇ Àǹ̸¦ Æ÷ÇÔÇÏ´Â ¹®Àå¼öÁý ¹æ¹ý |
¿µ¹®Á¦¸ñ(English Title) |
A Method for Collecting Sentences Entailing Ontology Schema Triple |
ÀúÀÚ(Author) |
Á¤Áø¿í
ÇÑ¿ëÁø
³ëűæ
¹Ú¼º¹è
ÀÌ»óÁ¶
¹Ú¼¼¿µ
Jin-Wook Jeong
Yong-Jin Han
Tae-Gil Noh
Seong-Bae Park
Sang-Jo Lee
Seyoung Park
|
¿ø¹®¼ö·Ïó(Citation) |
VOL 39 NO. 06 PP. 0488 ~ 0496 (2012. 06) |
Çѱ۳»¿ë (Korean Abstract) |
º» ³í¹®Àº ¿ÂÅç·ÎÁö ½ºÅ°¸¶ Æ®¸®ÇÃÀÇ Àǹ̸¦ Æ÷ÇÔÇÏ´Â ¹®Àå¼öÁý ¹æ¹ýÀ» Á¦¾ÈÇÑ´Ù. ¿ÂÅç·ÎÁö´Â ½ºÅ°¸¶ Æ®¸®ÇÃÀ» µû¸£´Â ´Ù·®ÀÇ ÀνºÅϽº Æ®¸®ÇÃÀÌ ÀÖ°í, À̵é ÀνºÅϽº Æ®¸®ÇÃÀº ½Ç¼¼°è¿¡¼ ÀÚ¿¬¾î ¹®Àå ÇüÅ·ΠµîÀåÇÒ ¼ö ÀÖ´Ù. º» ³í¹®¿¡¼´Â ÀÌ·¯ÇÑ ÀνºÅϽº Æ®¸®Çà Á¤º¸¸¦ È°¿ëÇÏ¿© ÁÖ¾îÁø ½ºÅ°¸¶ Æ®¸®ÇÃÀÇ Àǹ̸¦ Æ÷ÇÔÇÏ´Â ¹®ÀåÀ» ¼öÁýÇÑ´Ù. ÀνºÅϽº Æ®¸®ÇÃÀÌ ÀÚ¿¬¾î ¹®ÀåÀ¸·Î Ç¥ÇöµÉ ¶§ ƯÁ¤ Ç¥ÇöÀÌ ´Ù¾çÇÑ Àǹ̸¦ Æ÷ÇÔÇϰųª ÇϳªÀÇ Àǹ̰¡ ´Ù¾çÇÑ ÇüÅ·ΠǥÇöµÇ´Â ¸ðÈ£¼ºÀÌ ÀÖ´Ù. ÀÌ·¯ÇÑ ¸ðÈ£¼º ¹®Á¦¸¦ ÇØ°áÇϱâ À§ÇØ ¿ÂÅç·ÎÁö¿Í ¹®¼°£ÀÇ ¸í½ÃÀûÀÎ ¿¬°áÁ¤º¸¿Í ´Ù·®ÀÇ ÀνºÅϽº Æ®¸®Çà Á¤º¸¸¦ È°¿ëÇÑ´Ù. ½ÇÇè¿¡¼´Â DBpedia¸¦ È°¿ëÇÏ¿´À¸¸ç, Á¦¾ÈÇÏ´Â ¹æ¹ýÀÌ ±âÁ¸ ¿¬±¸¿¡ ºñÇØ ³ôÀº Á¤È®·ü°ú ÀçÇö·üÀ» º¸¿´´Ù. |
¿µ¹®³»¿ë (English Abstract) |
This paper proposes a method for collecting sentences which entail ontology schema triple. Many of instance triples conform to the schema triple of an ontology, these instance triples can present in the form of natural language sentences on the real world. In this paper, we use the information of instance triples for collecting sentences containing meaning of a schema triple. When instance triples are represented as the natural language form, there are several ambiguities which have multiple meanings, or the meaning of an expression appears in various forms. In order to solve these ambiguities, our method uses the explicit connection information between an entity of an instance triple and a document, and also makes full use of instance triples information. We experimented with the DBpedia. The result showed that as the instance triples increase, it was possible to collect the various predicate expressions and it showed the improved recall and precision compared with existing method of research. |
Å°¿öµå(Keyword) |
¹®Àå¼öÁý
ÄÚÆÛ½º ±¸Ãà
¿ÂÅç·ÎÁö ½ºÅ°¸¶ Æ®¸®ÇÃ
ÀÚ¿¬¾î ÀÎÅÍÆäÀ̽º
±×·¡ÇÁ ±¸Á¶Á¤º¸
Collecting Sentence
Corpus Construction
Ontology Schema Triple
Natural Language Interface
Graph Structure Information
|
ÆÄÀÏ÷ºÎ |
PDF ´Ù¿î·Îµå
|