Research on the classification and optimization strategies of civil aviation customer service based on BERTopic and the Kano model

Abstract

This study presents an innovative data-driven framework for optimizing airline customer service by integrating BERTopic topic modeling, SnowNLP sentiment analysis, and the Kano model. Unlike traditional approaches relying on surveys or subjective judgment, this method analyzes real customer service dialogues to extract 15 key service topics, assess satisfaction and attention scores, and categorize them into basic, expected, attractive, and indifferent needs. Findings show that Children's Ticket Inquiries, Pet Transportation, Baggage Regulations, and Seat Selection fall under basic needs, requiring prioritized investment to prevent dissatisfaction. Flight Rescheduling, Membership Verification, and Medical Refunds are expected needs that demand targeted improvements to enhance satisfaction. Promotional Inquiries and Expedited Services are attractive needs where innovative enhancements can create surprise and delight. Indifferent needs such as Standard Refunds, Meal Services, and Lost Item Handling require only baseline quality maintenance. Guided by the principle of demand-oriented resource allocation, the study proposes tailored optimization strategies for each category. This framework reveals latent customer priorities and transforms unstructured dialogue data into actionable insights, offering both theoretical contributions and practical implications for improving service quality and competitive positioning in the civil aviation sector.