• 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¸°úÇÐȸ³í¹®Áö (Journal of KIISE)

Á¤º¸°úÇÐȸ³í¹®Áö (Journal of KIISE)

Current Result Document :

ÇѱÛÁ¦¸ñ(Korean Title) ´º·²-½Éº¼¸¯ ¼øÀ§È­ ¸ðµ¨ ±â¹Ý 2´Ü°è ´Ü¶ô Àç¼øÀ§È­ ¸ðµ¨
¿µ¹®Á¦¸ñ(English Title) 2-Phase Passage Re-ranking Model based on Neural-Symbolic Ranking Models
ÀúÀÚ(Author) ¹è¿ëÁø   ±è Çö   ÀÓÁØÈ£   ±èÇö±â   ÀÌ°øÁÖ   Yongjin Bae   Hyun Kim   Joon-Ho Lim   Hyun-ki Kim   Kong Joo Lee  
¿ø¹®¼ö·Ïó(Citation) VOL 48 NO. 05 PP. 0501 ~ 0509 (2021. 05)
Çѱ۳»¿ë
(Korean Abstract)
ÀÚ¿¬¾î ÁúÀÇÀÀ´ä ½Ã½ºÅÛ°ú °ü·ÃÇÑ ÀÌÀüÀÇ ¿¬±¸µéÀº ÁÖ¾îÁø Áú¹®°ú ´Ü¶ôÀ¸·ÎºÎÅÍ Á¤È®ÇÑ Á¤´äÀ» ÃßÃâÇÏ´Â ¹®Á¦¿¡ ÃÊÁ¡À» ¸ÂÃß°í ÀÖ´Ù. ±×·¯³ª, ±â°èµ¶ÇØ¿¡¼­ ¿ÀÇ µµ¸ÞÀÎ ÁúÀÇÀÀ´äÀ¸·Î ¹®Á¦¸¦ È®ÀåÇÏ¿´À» ¶§, Á¤´äÀÌ Æ÷ÇÔµÈ ´Ü¶ôÀ» Àß Ã£´Â °ÍÀÌ ±â°èµ¶ÇØ ¸øÁö¾ÊÀº Áß¿äÇÑ ¿ä¼ÒÀÌ´Ù. DrQA[1]¿¡¼­´Â Ãʱ⠰˻ö ´Ü°è¸¦ Æ÷ÇÔÇÏ¿© ÁúÀÇÀÀ´äÀ» ÇÏ¿´À» ¶§ Exact Match@Top1 ¼º´ÉÀÌ 69.5%¿¡¼­ 27.1%·Î Ç϶ôÇß´Ù°í Æò°¡ÇÏ¿´´Ù. º» ³í¹®¿¡¼­´Â ÁúÀÇÀÀ´ä ½Ã½ºÅÛ ¼º´É Çâ»óÀ» À§ÇØ 2´Ü°è ´Ü¶ô Àç¼øÀ§È­ ¸ðµ¨À» Á¦¾ÈÇÑ´Ù. 2´Ü°è ´Ü¶ô Àç¼øÀ§È­ ¸ðµ¨Àº ½Éº¼¸¯ ¼øÀ§È­ ¸ðµ¨°ú ´º·² ¼øÀ§È­ ¸ðµ¨ÀÇ °á°ú¸¦ ÅëÇÕÇÏ¿© ´Ù½Ã Àç¼øÀ§È­ÇÏ´Â ¸ðµ¨ÀÌ´Ù. ½Éº¼¸¯ ¼øÀ§È­ ¸ðµ¨Àº CatBoost ¾Ë°í¸®Áò°ú Áú¹®°ú ´Ü¶ô °£ÀÇ ÀÚÁúÀ» ±â¹ÝÀ¸·Î ´Ü¶ôÀ» ¼øÀ§È­ÇÏ°í, ´º·² ¼øÀ§È­ ¸ðµ¨Àº Çѱ¹¾î µö·¯´× ¾ð¾î¸ðµ¨(KorBERT)À» »çÈÄÇнÀÇÏ¿© ¼øÀ§È­ÇÏ¿´´Ù. 2´Ü°è ¸ðµ¨Àº ´º·² ¸®±×·¹¼Ç ¸ðµ¨¿¡ ±â¹ÝÇÏ¿© ¼øÀ§È­ÇÏ¿´´Ù. º» ³í¹®¿¡¼­´Â Ư¡ÀÌ ´Ù¸¥ ¼øÀ§È­ ¸ðµ¨À» °áÇÕÇÏ¿© ¼º´ÉÀ» ±Ø´ëÈ­ÇÏ¿´°í, ÃÖÁ¾ÀûÀ¸·Î Á¦¾ÈÇÑ ¸ðµ¨Àº 1,000°ÇÀÇ Áú¹®À» Æò°¡ÇÏ¿´À» ¶§ MRR ±âÁØ 85.8%°ú BinaryRecall@Top1±âÁØ 82.2%ÀÇ ¼º´ÉÀ» º¸¿´°í, °¢ ¼º´ÉÀº º£À̽º¶óÀÎ ¸ðµ¨º¸´Ù 17.3%(MRR), 22.3%(BR@Top1)ÀÌ Çâ»óµÇ¾ú´Ù
¿µ¹®³»¿ë
(English Abstract)
Previous researches related to the QA system have focused on extracting exact answers for the given questions and passages. However, when expanding the problem from machine reading comprehension to open domain question answering, finding the passage containing the correct answer is as important as machine reading comprehension. DrQA[1] reported that Exact Match@Top1 performance decreased from 69.5 to 27.1 when the QA system had the initial search step. In the present work, we have proposed the 2-phase passage reranking model to improve the performance of the question answering system. The proposed model integrates the results of the symbolic and neural ranking models to re-rank them again. The symbolic ranking model was trained based on the CatBoost algorithm and manual features between the question and passage. The neural model was trained based on the KorBERT model by fine-tuning. The second stage model was trained based on the neural regression model. We maximized the performance by combining ranking models with different characters. Finally, the proposed model showed the performance of 85.8% via MRR and 82.2% via BinaryRecall@Top1 measure while evaluating 1,000 questions. Each performance was improved by 17.3%(MRR) and 22.3%(BR@Top1) compared with the baseline model.
Å°¿öµå(Keyword) ´Ü¶ô Àç¼øÀ§È­   Á¤º¸°Ë»ö   ÁúÀÇÀÀ´ä   Á¤º¸ÃßÃâ   passage re-ranking   information retrieval   question answering system   information extraction  
ÆÄÀÏ÷ºÎ PDF ´Ù¿î·Îµå