Comparative analysis of changes in passenger's perception for airline companies' service quality before and during COVID-19 using topic modeling

The purpose of this paper is to compare changes in air passenger's perceptions of airline services before and during COVID-19. To conduct the comparative analysis, two approaches have been employed, topic modeling and sentimental analysis. Firstly, influential topics that reflect passenger's opinions are extracted from online reviews using the Latent Dirichlet Allocation (LDA) model. These topics are then distributed into the five dimensions of the SERVQUAL model to analyze the changes in importance of each service quality dimension. Moreover, online reviews are classified as negative or positive based on the review score. Using logistics regression with negative reviews, this paper analyzes which keywords affect passenger dissatisfaction with airline services and compares the changes in negative keywords before and during COVID-19. The results show that the reliability of airlines was the most significant dimension before COVID-19, while the responsiveness of airlines has become the most crucial dimension during COVID-19. From the results of the sentimental analysis, words such as ‘pay', ‘refund', ‘book', and ‘cancelled' can be identified as keywords that bring about negative emotions, indicating that airline companies need to focus upon refund and reservation services to enhance their service level. Based upon this analysis, the paper makes several suggestions for improving airline service quality in light of the changes in air passenger perception.