Loading...
Paper Type
Complete
Description
The Consumer Financial Protection Bureau (CFPB) receives and processes consumer complaints about various financial services. Every complaint provides insight into problems that consumers experience. With an increasing number of CFPB complaints, manual prioritisation of these complaints is not feasible. This paper proposes a smart approach based on latent Dirichlet allocation (LDA) and Emotion detection to analyse the CFPB consumer complaints, extracting latent topics and emotions from CFPB complaints and assigning priority based on a predefined priority matrix. The automated priority recommendation engine provides faster and optimum prioritisation. The use of emotion detection and LDA allows categorization of complaints using two distinct aspects, topic, and emotion, providing varied priorities for the complaints. The average weighted priority calculation based on topics and emotions allows the recommendation engine to remove the dependence of prioritisation systems on user-selected labels. The proposed approach allows a configurable matrix, making the system implementation easy to configure as per business requirements.
Paper Number
1933
Recommended Citation
Marshan, Alaa; Joshi, Nikhil; Ioannou, Athina; and Spanaki, Konstantina, "Customer Complaints Response Priority Recommendation Engine: A HYBRID NLP Model" (2023). AMCIS 2023 Proceedings. 17.
https://aisel.aisnet.org/amcis2023/sig_dsa/sig_dsa/17
Customer Complaints Response Priority Recommendation Engine: A HYBRID NLP Model
The Consumer Financial Protection Bureau (CFPB) receives and processes consumer complaints about various financial services. Every complaint provides insight into problems that consumers experience. With an increasing number of CFPB complaints, manual prioritisation of these complaints is not feasible. This paper proposes a smart approach based on latent Dirichlet allocation (LDA) and Emotion detection to analyse the CFPB consumer complaints, extracting latent topics and emotions from CFPB complaints and assigning priority based on a predefined priority matrix. The automated priority recommendation engine provides faster and optimum prioritisation. The use of emotion detection and LDA allows categorization of complaints using two distinct aspects, topic, and emotion, providing varied priorities for the complaints. The average weighted priority calculation based on topics and emotions allows the recommendation engine to remove the dependence of prioritisation systems on user-selected labels. The proposed approach allows a configurable matrix, making the system implementation easy to configure as per business requirements.
When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.
Comments
SIG DSA