Supporting sustainable system adoption: Socio-semantic analysis of transit rider debates on social media

Document Type

Journal Article

Publication Date

2018

Subject Area

place - north america, planning - surveys, planning - signage/information, technology - passenger information

Keywords

Customer satisfaction, Energy saving, Ridership behavior, Social media, Social network analysis, Semantic analysis

Abstract

Online social media platforms provide a bi-directional communication channel between transit agencies and their customers. It can also be an effective venue for profiling users and their needs as a step towards customizing service and communication policies. This study analyzed online Twitter discussions for three major transit agencies in Canada. In our work, we integrate the analysis of the participants’ social networks with the contents of their discussion. We also conduct the semantic analysis in a manner that parallels the structure and contents of customer satisfaction surveys—allowing for insightful comparisons between the results of both methods Analysis of the structure of the social networks of the Twitter accounts under study, including investigation of the formation of sub-communities and their interrelationship to the overall network. It was found that networks of the three cases portray a scale free and small world behaviors. This means that they are maturing networks; and that they represent viable communities—not just a randomly connected graph. This is important for future studies in relation to information diffusion and opinion dynamics: how people share information and how does this help shape their views. On the semantic front, a lexicon was developed based on existing thesauri for customer satisfaction analysis. Keywords form each tweet were extracted and the topic(s) of each tweet was defined based on the lexicon. It was found that, based on the sample investigated, the behavior of 100-follower networks (networks with nodes having at least 100 followers) closely mimics the behavior of the overall network. Studying these networks (of influential users) can make analysis faster and may not impact accuracy. We also clustered each network into sub-networks: small, medium, and large. Topics discussed in medium-size networks tended to be unique. This seems to be the level where active discussions of specific topics take place. Focusing on detecting these and analyzing their contents can provide better chance for capturing the evolution of community opinions.

Rights

Permission to publish the abstract has been given by Elsevier, copyright remains with them.

Comments

Sustainable Cities and Society

http://www.sciencedirect.com/science/journal/22106707/22/supp/C

Link to Full Text

Share

COinS