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

»çÀÌÆ®¸Ê

Loading..

Please wait....

±¹³» ÇÐȸÁö

Ȩ Ȩ > ¿¬±¸¹®Çå > ±¹³» ÇÐȸÁö > µ¥ÀÌÅͺ£À̽º ¿¬±¸È¸Áö(SIGDB)

µ¥ÀÌÅͺ£À̽º ¿¬±¸È¸Áö(SIGDB)

Current Result Document : 23 / 23

ÇѱÛÁ¦¸ñ(Korean Title) ºñ ÀÛ¾÷ÁöÇâ 꺿 ¸ðµ¨ÀÇ ½ÇÇèÀû ºñ±³ ¿¬±¸
¿µ¹®Á¦¸ñ(English Title) An Empirical Comparison Study of Non-Task-Oriented Chatbot Models
ÀúÀÚ(Author) Á¶¸¸Àç   ¿À³²ÈÆ   À̱âÈÆ   Man-Jae Cho   Nam-Hun Oh   Ki-Hoon Lee  
¿ø¹®¼ö·Ïó(Citation) VOL 35 NO. 02 PP. 0003 ~ 0015 (2019. 08)
Çѱ۳»¿ë
(Korean Abstract)
Ãֱ٠꺿¿¡ ´ëÇÑ ¿¬±¸°¡ È°¹ßÈ÷ ÁøÇàµÇ°í ÀÖ´Ù. 꺿Àº ÀÛ¾÷ÁöÇâ 꺿°ú ºñ ÀÛ¾÷ÁöÇâ 꺿À¸·Î ³ª´­ ¼ö ÀÖ´Ù. »ç¶÷°ú ´Ù¾çÇÑ ÁÖÁ¦ÀÇ ´ëÈ­¸¦ ³ª´­ ¼ö ÀÖ´Â ºñ ÀÛ¾÷ÁöÇâ 꺿Àº ÁÖ·Î ½ÃÄö½º Åõ ½ÃÄö½º (Sequence-to-Sequence) ¸ðµ¨À» ±â¹ÝÀ¸·Î ÇÑ´Ù. º» ³í¹®¿¡¼­´Â ´Ù¾çÇÑ ½ÃÄö½º Åõ ½ÃÄö½º ±â¹Ý ¸ðµ¨À» ºñ ÀÛ¾÷ÁöÇâ 꺿¿¡ Àû¿ëÇغ¸°í ±× ¼º´ÉÀ» ½ÇÇèÀûÀ¸·Î ºñ±³ÇÑ´Ù. ¿µÈ­ ´ë»ç µ¥ÀÌÅÍ ÁýÇÕÀ» »ç¿ëÇÏ¿© ½ÇÇèÇÑ °á°ú, Æ®·£½ºÆ÷¸Ó ¸ðµ¨ÀÌ Á¤·® ¹× Á¤¼ºÆò°¡¿¡¼­ °¡Àå ¿ì¼öÇÑ ¼º´ÉÀ» º¸¿´´Ù. ¶ÇÇÑ, ÇнÀ µ¥ÀÌÅÍ¿¡ ±ºÁýÈ­¸¦ Àû¿ëÇϸé BLEU Á¡¼ö°¡ Çâ»óµÇ´Â ¹Ý¸é ´Ù¾ç¼º Á¡¼ö´Â ÀúÇϵǴ °ÍÀ» ¹ß°ßÇÏ¿´´Ù.
¿µ¹®³»¿ë
(English Abstract)
Recently, there have been active research efforts on chatbots. Chatbots can be divided into task-oriented and non-task-oriented chatbots. Non-task-oriented chatbots, which can converse with humans on various topics, are mainly based on Sequence-to-Sequence models. In this paper, we apply various Sequence-to-Sequence models to non-task-oriented chatbots and conduct an empirical comparison study. Experimental results using a movie dialogue dataset show that the Transformer model is the best one in quantitative and qualitative evaluation. We also discovered that applying clustering to the train data increases the BLEU score, but decreases the diversity score.
Å°¿öµå(Keyword) 꺿   ½ÃÄö½º Åõ ½ÃÄö½º   ÁÖÀÇÁýÁß ±â¹ý   Æ®·£½ºÆ÷¸Ó   °­È­ÇнÀ   ±ºÁýÈ­   Chatbot   Sequence-to-Sequence   Attention Mechanism   Transformer   Reinforcement Learning   Clustering  
ÆÄÀÏ÷ºÎ PDF ´Ù¿î·Îµå