Classification of Service Types from Customer Feedback Using Deep Learning Methods

Author
Pongsuda Banchuen, Sooksawaddee Nattawuttisit
Abstract
This paper proposes a classification model using Long Short-Term Memory of Recurrent Neural Network (LSTM) for the customer feedback. To evaluate the customer feedback, the customer relationship management (CRM) system and the dataset of 275,670 responses in transportation sector are also used in this research. The collected customer feedback can be classified into two categories which are complaints and commendations. To build the LSTM model using dataset, we provide 192,970 samples for training dataset, 41,350 samples for validation dataset, and 41,350 samples for test dataset. After the model’s evaluations, the performance results of LSTM model in terms of accuracy, precision, recall, and f-measure are 99.04%, 96.73%, 90.12%, and 93.15%, respectively. In the same way, the performance results of bidirectional LSTM (Bi-LSTM) model in terms of accuracy, precision, recall, and f-measure are 99.05%, 97.00%, 90.13%, and 93.27% respectively. It is clear that the both of system models are comparable and could be used for CRM.
Keywords
Neural Network, Deep Learning, Customer Services