Utilizing Natural Language Processing in Chatbots for Integrated Complaint Management Application: A Case Study of PT PLN (Persero) ULP Padalarang

Customer service operations at PT PLN (Persero), Indonesia’s main electricity provider, face considerable challenges in effectively managing customer inquiries. Despite establishing key services through the PLN 123 Contact Center and a chat feature in the PLN mobile app, the current system leads to a backlog of complaints, resulting in delayed responses to customer issues and inadequate resolutions. This research aims to develop a tailored intent classification model for a customer service chatbot within the PLN Mobile application using a transformer-based embedding approach, specifically IndoBERT. A comparative analysis of the transformer model was performed through various experiments focusing on Learning Rate and Optimizer parameters. The results consistently demonstrate the effectiveness of the proposed model, achieving with an accuracy of 93%, precision of 90%, recall of 93%, and an F1 score of 91% and evaluation loss of 0.658680. Thus, the IndoBERT model exhibits strong performance in enhancing customer service chatbots, particularly for general inquiries regarding electricity problems and the PLN Mobile application.