Abstract
Complaint handling system used by financial companies are handled by live agents these days, there’s a need to move from a system handled by live agents to a system which automatically handles the complaints to increase efficiency & save cost & time. We are planning to develop an automatic financial complaint classification system that automatically deals with the customer complaints by segregating the data & routing it to the right department. We are planning to develop the system by using Natural Language Processing (NLP), Artificial Intelligence (AI), Machine Learning (ML) & Deep Learning (DL) concepts and implement using Python, Jupyter Notebook,.etc. The end product will be a web- based application system where customer can register their complaints with- out having to worry about sending it to right department. (Bejarano) Devel- oped system will automatically segregate the complaints & route it to the right department. Through this project we are trying to attain best results for our complaint classification task by comparing various Machine Learning (ML) models, Deep Learning (DL) models and Ensemble methods on basis of accu- racy and time and applying the one which best suits the requirement. (Zhang, Zhao, and Lecun) We are using data pre-processing methods like data augmen- tation, lemmatization etc and on top of that TF-IDF and Word2Vec methods for ML and DL models respectively.
)