2019³âµµ ÀÎÅͳÝÁ¤º¸ÇÐȸ Ãá°èÇмú¹ßÇ¥´ëȸ
ÇѱÛÁ¦¸ñ(Korean Title) |
SVM based intent, entity detection System |
¿µ¹®Á¦¸ñ(English Title) |
SVM based intent, entity detection System |
ÀúÀÚ(Author) |
Sudan Prasad Uprety
HyeonWook Kim
Seung Ryul Jeong
|
¿ø¹®¼ö·Ïó(Citation) |
VOL 20 NO. 01 PP. 0167 ~ 0168 (2019. 04) |
Çѱ۳»¿ë (Korean Abstract) |
|
¿µ¹®³»¿ë (English Abstract) |
With the rapid development of technology and machine learning technic, Speech understanding and Dialog Systems are now a global trend and has drawn attention from both academic and business communities. Although the Deep learning technology is getting more interest from academic and business sector, traditional machine learning algorithms such as SVM, Bayesian network are also getting equal interest in dialogue system. Support vector machine can detect intent or extract entity separately. In this paper we focus on SVM based joint approach to predict intent and entity simultaneously to improve prediction accuracy in Dialog Systems. With the help of joint model down-stream error, i.e. predicting one based on other can be reduced and increase the accuracy on intent and entity prediction.
|
Å°¿öµå(Keyword) |
Chatbot
Natural Language Processing
Support Vector Machine
Joint Model
|
ÆÄÀÏ÷ºÎ |
PDF ´Ù¿î·Îµå
|