Towards the Ontology Development for Smart Transportation Infrastructure Planning via Topic Modeling

Media age has seen a huge amount of data flowing in from all directions, be it online news sources, social media, technical documents, and many more. There is a huge scope of these data sources for utilization in the transportation sector that can potentially improve the current practice of transportation infrastructure planning. In order to effectively capture, analyze, and utilize the information from various sources, ontologies are useful tools as they can provide clear and structured knowledge in the transportation domain. Majority of the existing transportation-related ontologies focus on traffic management and route planning. The objective of this paper is to initiate the development of an integrated ontology that can help with long-term planning and decision-making of transportation infrastructure by proposing a preliminary taxonomy in this domain. To this end, 20 transportation planning visionary documents published by government agencies were collected and analyzed using topic modelling techniques. Specifically, two topic modeling methods: Latent Dirichlet Allocation (LDA) and Non-negative Matrix Factorization (NMF) models were used to extract important and emerging concepts related to transportation infrastructure planning. Leveraging the important and emerging concepts, a preliminary taxonomy of transportation infrastructure planning was then developed and presented.