고객 서비스 개선을 위한 비정형 텍스트 데이터 자동 분류 모델 비교 - 고객 불만 데이터를 대상으로

Abstract

With the advancement of information technology and the proliferation of digital communication, there is an increasing need for effective management and processing of the growing volume of customer complaints and feedback. Particularly, when most of the data is unstructured, it is crucial to quickly process and classify this data to improve customer service. In this study, we examine traditional classification models such as Naive Bayes, SVM, and Random Forest, as well as deep learning models like CNN and LSTM. These models were applied to Boston's traffic complaint data to compare the performance of automatic classification of customer complaint data. The results showed that CNN and LSTM achieved high classification accuracies of 81% and 97%, respectively, confirming their effectiveness in handling complex and diverse patterns of unstructured data, such as customer complaints. These findings demonstrate that deep learning models are better at analyzing the characteristics of customer complaint data and detecting contextual connections. This study provides a methodology for companies to quickly and accurately understand and respond to customer voices, thereby enhancing corporate competitiveness.