Framework for evaluating online public opinions on urban rail transit services through social media data classification and mining

Abstract

Urban rail transit (URT) service quality assessments are pivotal for transport authorities to gauge passenger preferences and refine operational strategies. Online public opinion offers a vast pool of data at a reduced acquisition cost compared to traditional survey methods. However, current research lacks effective methodologies for classifying and interpreting extensive social media data (SMD) related to URT services. This study presents a comprehensive framework tailored to efficiently classify and mine public opinion on URT services from social media platforms. Leveraging data from ten Chinese cities with extensive URT networks, a domain-specific lexicon is semi-automatically constructed by integrating official documents (standards, policies, and annual reports) and high-frequency online terms. Additionally, a text classification algorithm based on this lexicon is proposed. Subsequently, sentiment, semantic, and timeline analyses are conducted on the classified texts to extract public opinion. Importantly, many manual steps employed in this study can be avoided when extended to other application scenarios. Therefore, this study contributes to the advancement of SMD processing efficiency in the URT domain and holds promise for broader applications in the fields of transportation management and policy-making.