The Effectiveness of Sentiment Analysis for Detecting Fine-grained Service Quality

Improving public transport relies on the development of appropriate tools for measuring and monitoring service quality. While there are several studies exploring perceived service quality, these works are mainly based on surveys, which are limited to a pre-de fined period and do not enable to constantly observe and analyze the respond of passengers to different events. As a result, they may miss events affecting service quality in a transport hub. In contrast, social media feeds provide an opportunity to constantly look for events that may affect the service quality. However, in a confi ned geographic area like a transport hub, the sparsity of data considerably reduce the effectiveness of straight-forward event detection methods. In contrast
to earlier works and in order to face these challenges, in this paper, we explore the effectiveness of sentiment analysis to detect events impacting perceived service quality in a con fined location. Southern Cross Station, a major transport hub in Melbourne, Australia, is selected as a case
study. Moreover, a statistical approach to combine and integrate sentiment-based and frequency-based
event detection is proposed. Main findings con rm the effectiveness of sentiment analysis for detecting events affecting service quality in a transport hub. Additionally, the results prove that the proposed method can improve the sensitivity of event detection for events affecting
service quality.