• 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¸°úÇÐȸ ÄÄÇ»ÆÃÀÇ ½ÇÁ¦ ³í¹®Áö (KIISE Transactions on Computing Practices)

Á¤º¸°úÇÐȸ ÄÄÇ»ÆÃÀÇ ½ÇÁ¦ ³í¹®Áö (KIISE Transactions on Computing Practices)

Current Result Document : 1 / 1

ÇѱÛÁ¦¸ñ(Korean Title) µö·¯´×À» È°¿ëÇÑ ±ÝÀ¶»óÇ° ºÒ¿ÏÀü ÆǸŠÆǺ° ¸ðµ¨
¿µ¹®Á¦¸ñ(English Title) Discrimination Model On Misselling of Financial Products Using Deep Learning
ÀúÀÚ(Author) ±èÁÖÇö   ¿øÁ¤ÀÓ   JuHyun Kim   Jung-Im Won  
¿ø¹®¼ö·Ïó(Citation) VOL 25 NO. 06 PP. 0294 ~ 0302 (2019. 06)
Çѱ۳»¿ë
(Korean Abstract)
Æݵå·Î ´ëÇ¥µÇ´Â ±ÝÀ¶»óÇ°Àº ÆǸÅÁ÷¿øÀÇ ¼³¸í°ú ±ÇÀ¯°¡ Àý´ëÀûÀÎ ÈûÀ» ¹ßÈÖÇÏ´Â ±¸Á¶Àû Ư¼ºÀ¸·Î ÀÎÇØ ºÒ¿ÏÀüÆǸŠ°¡´É¼ºÀ» ³»Æ÷ÇÏ°í ÀÖÀ¸¸ç, ÀÌ·Î ÀÎÇÑ »çȸÀû °¥µîÀÌ ²ÙÁØÈ÷ ÃÊ·¡µÇ°í ÀÖ´Ù. º» ³í¹®¿¡¼­´Â ÅؽºÆ®·Î º¯È¯µÈ »ó´ã µ¥ÀÌÅÍ¿¡ µö·¯´× ±â¹Ý ÅؽºÆ® ºÐ·ù ±â¹ýÀ» Àû¿ëÇØ ±ÝÀ¶»óÇ° ºÒ¿ÏÀüÆǸŠ¿©ºÎ¸¦ ÆǺ°ÇÏ´Â ¸ðµ¨À» Á¦¾ÈÇÑ´Ù. º» ¸ðµ¨Àº ÇѱÛÀÇ ÀÚ¼Ò ´ÜÀ§ÀÇ ÅäÅ«È­ ¹æ¹ýÀ» Àû¿ëÇÑ º¤ÅÍÈ­¿Í ±âÁ¸ ÄÁº¼·ç¼Ç°ú ¼øȯ ½Å°æ¸ÁÀ» ÀÌ¿ëÇÑ ºÐ·ù ¸ðµ¨À» ±â¹ÝÀ¸·Î ÇÑ´Ù. ƯÈ÷, µÎ ¸íÀÇ È­ÀÚ°¡ Á¸ÀçÇÏ´Â »ó´ã µ¥ÀÌÅÍÀÇ Æ¯¼ºÀ» °¨¾ÈÇØ È­ÀÚ º¤Å͸¦ Ãß°¡ °í·ÁÇÔÀ¸·Î½á, ÀϹÝÀûÀÎ µö·¯´× ÅؽºÆ® ºÐ·ù ±â¹ý º¸´Ù ¸Å¿ì ÁÁÀº ¼º´ÉÀ» º¸ÀÓÀ» È®ÀÎ ÇÏ¿´´Ù. ¶ÇÇÑ ½ÇÁ¦ ±ÝÀ¶ »ê¾÷¿¡¼­ ºÒ¿ÏÀüÆǸŠ¿©ºÎ¸¦ ÆǺ°Çϱâ À§ÇÏ¿© »ç¿ëµÇ´Â ½Ã³ª¸®¿À¸¦ Åä´ë·Î »ý¼ºµÈ °¡°øÀÇ µ¥ÀÌÅ͸¦ »ç¿ëÇÑ ½ÇÇè °á°ú¸¦ Á¦½ÃÇÏ¿© º» ¸ðµ¨ÀÇ ½ÇÈ¿¼ºÀ» °ËÁõÇÏ¿´´Ù.
¿µ¹®³»¿ë
(English Abstract)
Financial products represented by funds have structural characteristics, of which explanation and recommendation by sales associates, exert a major influence on customers¡¯ purchasing decisions. So, there always exists the possibility of associates misselling financial products, resulting in chronic social conflicts between sales associates and customers. This paper proposes the discrimination model on misselling of financial products, using deep learning-based text classification, for converted customers¡¯ call text data. The model is based on vectorization, which applies Hangeul¡¯s character-level tokenization, and a classification method using CNN and RNN. Specifically, we have validated that the model shows much better performance, by including the speaker vector because there are two speakers in real customers¡¯ call data. Also, we have validated effectiveness of the proposed model, by presenting experimental results which use generated synthetic data. Such data is based on real business cases, that are used to prevent misselling of financial products in the financial industry.
Å°¿öµå(Keyword) ±ÝÀ¶»óÇ° ºÒ¿ÏÀü ÆǸŠ  µö·¯´×   ÄÁº¼·ç¼Ç ½Å°æ¸Á   ¼øȯ ½Å°æ¸Á   misselling of financial products   deep learning   CNN   RNN  
ÆÄÀÏ÷ºÎ PDF ´Ù¿î·Îµå