TY - GEN
T1 - Utilizing Unsupervised kNN on Customer Risk Assessment for Banking
AU - Huang, Ching Jung
AU - Chien, Kuan Chen
AU - Fang, Shen Wei
AU - Huang, Chun Hua
AU - Chen, Shan Yi
AU - Chang, Yu Ping
AU - Teng, Wei Guang
N1 - Funding Information:
ACKNOWLEDGMENT This study was supported in part by Bank Sinopac and by Atelier Future in National Cheng Kung University, Taiwan, under contracts B109-K550A.
Publisher Copyright:
© 2022 IEEE.
PY - 2022
Y1 - 2022
N2 - Facing the rise of anti-money laundering and anticapital terrorism, financial institutions spend a lot of money and time every year to conduct Know Your Customer (KYC) verification for customers. Therefore, we devise an approach to carefully handle the tasks of data processing. Specifically, we choose to adopt instance-based learning in this work to consider each target customer one at a time. Also, we separate all the data features into profile and transaction ones. Based on the concept that customers of similar profiles tend to have similar transaction behaviors, we calculate the outlier score of each customer. Only when a suspicious customer having a high outlier score is determined, further actions of manual screening should be taken. Moreover, our approach is explainable as statistics of the corresponding profile and transaction features are reported. With these carefully designed means, our approach helps to significantly improve and speed up the whole KYC process.
AB - Facing the rise of anti-money laundering and anticapital terrorism, financial institutions spend a lot of money and time every year to conduct Know Your Customer (KYC) verification for customers. Therefore, we devise an approach to carefully handle the tasks of data processing. Specifically, we choose to adopt instance-based learning in this work to consider each target customer one at a time. Also, we separate all the data features into profile and transaction ones. Based on the concept that customers of similar profiles tend to have similar transaction behaviors, we calculate the outlier score of each customer. Only when a suspicious customer having a high outlier score is determined, further actions of manual screening should be taken. Moreover, our approach is explainable as statistics of the corresponding profile and transaction features are reported. With these carefully designed means, our approach helps to significantly improve and speed up the whole KYC process.
UR - http://www.scopus.com/inward/record.url?scp=85151977120&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85151977120&partnerID=8YFLogxK
U2 - 10.1109/ICSTE57415.2022.00022
DO - 10.1109/ICSTE57415.2022.00022
M3 - Conference contribution
AN - SCOPUS:85151977120
T3 - Proceedings - 2022 12th International Conference on Software Technology and Engineering, ICSTE 2022
SP - 103
EP - 108
BT - Proceedings - 2022 12th International Conference on Software Technology and Engineering, ICSTE 2022
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 12th International Conference on Software Technology and Engineering, ICSTE 2022
Y2 - 25 October 2022 through 27 October 2022
ER -
